#chatgpt-discussions

1 messages · Page 75 of 1

summer pagoda
#

history is not neccecarely bad, but outdated information to lets say population density and other cruical stuff is defenetely bad

tight island
#

History is important to teach

#

They’re just too selective about it

summer pagoda
#

yes. Mostly tho its closely linked to the countrys history itself. Romans, WW1/2 and other topics get more likely teached in european schools vs others

tight island
#

Heck I think the current method of schooling needs a major overhaul

#

6-9 hours of school?, are you insane

summer pagoda
#

One thing i would agree, is to learn how to think and not to learn how to remember "stuff". Tho lastly has its own benefits of course. Its difficult topic

tight island
#

It is, we still use the factory model

summer pagoda
#

factory model?

tight island
#

Sit Down don’t speak raise your hand if you want something etc etc

summer pagoda
#

tho i must say, it does say ```
“But the magnitude of time saved and quality increases are very large in our paper, so it does seem like this is pretty revolutionary, at least for certain types of work.”

so as previously mentioned it depends in which field XD
summer pagoda
tight island
#

Originally a punishment

summer pagoda
#

Why a punishment?

tight island
#

Idk ask the creator of homework

summer pagoda
#

perhaps its not a punishment. I am sometimes a good orator and notice people listen in silence if they love to listen to my dialogue. So beeing quite is also a way to follow up on a topic

tight island
#

A classroom has more benefit to function the way “Ad Astra” does

tight island
#

Let kids be kids in their free time

severe summit
#

ok i wrote 2 - 3 messages with GPT4 and its limited now, i must wait 2 hours

summer pagoda
#

but i still dont understand how you can have an orator speaking while everyone else is talking. how can people follow up on the topic?

tight island
#

If you have classrooms be directed based on interest instead of orders from above it would help

#

I do know that the teachers hands are tied

summer pagoda
#

my sister is interested in japanese love movies and korean music bands but not about history. When i told her about caesar and the triumphirates she first was not interested. but when she had a test at school, i had to tell her everything around 50bce and we have a conversation for about 2 hours. She was in the end fascinated about it.

nevertheless she never thought of history as that cool until i took my time having a personal conversation with it. I doubt teachers have the time to do so with each individual student

tight island
#

Everyone I knew saw homework as useless and saw it as a punishment

#

Kids at grade 0 in my country has to be In the classroom 6 hours they have two breaks one 15 minutes and one 45 minutes

#

That’s ludicrous

#

Plus they get homework

summer pagoda
#

sit down with your kid, find a topic you both have interest and help the teachers out in this regard.
As Cato the elder loved to say "carthago delenda est" after each speach in the senate. In the final end after 50 years he got his wish. He kept it in peoples mind, and kept reminding them

tight island
#

The older ones are there 9 hours with the same breaks and have the same amount of breaks plus a. Enormous amount of homework

dense halo
#

I asked chatgpt, and she said it's ok for me to use it the way I want to and it's not plagarism since it's my thoughts anyway.

tight island
#

Homework belongs in school and not in the home or their spare time if they don’t want to

summer pagoda
#

You bring me to an obscure idea. ChatGpt is a replacement for the teacher to sit down and explain everything "in person". This is what i think kids realy need

tight island
#

I personally think that each individual assignment should be custom tailored to the interests of the students

#

No more uniform tests

#

Down to the individual student

summer pagoda
#

They unforunately dont have time, also we already have not enought teachers in switzerland

tight island
#

That’s where AI comes in

summer pagoda
#

pretty much. unfortunately kids yet have to also to learn to ask the right question. Think still the best approach is to spend time kids helping them out in the first place. ChatGpt cant "think" for you

tight island
#

Like a kid wants to be an astronaut give him tests based on that which includes what you must teach

summer pagoda
#

You most likely shatter the kids dream. They yet have to learn what it takes to be an austronaut. And its quite a lot...

tight island
#

Sure how long if til the earth rotates etc, with how to solve it shown exactly how

#

Issue I’ve noticed is it’s a one size fits all

#

I never did any homework, not even in the classroom I still score A+ because all I had to do is repeat then forget

summer pagoda
#

I think better show them different topics from school in real live and follow up if they show interest in one of the fields.

tight island
#

If I personally wanted to learn something I sought it out

summer pagoda
tight island
#

I agree

#

I’m just saying that the one size fits all sitting down listening to someone speak about a useless subject which they can’t use in the private life or even in the future

#

That needs to go

summer pagoda
#

But i have to stop now. i am still at work and need to focus on some problem solving XD
If you have an interesting topic in this field to read through, drop me some context i can google for 🙂

tight island
#

Can’t stop won’t stop

#

Same I have to stop as well otherwise Lugui is going to scold me

winged bluff
#

Is chatgpt down since for some reason its very slow and its not showing up when i refreshed the page.or not responding to any prompts

maiden fractal
inland heron
winged bluff
maiden fractal
winged bluff
#

Only if the creator s notice it not true

prime nymph
#

Does anyone have a model, library, or code for converting handwritten text to text/PDF? I would be so grateful for any assistance.

uncut creek
#

guys chatgpt keeps asking me to verify im not a robot

#

it isnt letter me in, yall know what's wrong

inland heron
uncut creek
#

😭

maiden fractal
uncut creek
#

i just noticed that and it worked after i switched locations

maiden fractal
full veldt
#

Have the GPT4 limits been lowered today? I seem to be hitting the limit much more quickly than usual

inland heron
#

but I'm not sure if it's still like this

proven lagoon
#

I hit my limit after 20 messages twice today. So yeah...

proud matrix
#

Oh cool it's not just me

#

I guess it's time to start using one of the competitors

fluid skiff
#

are we really just guessing at something we pay for?

proud matrix
fluid skiff
#

cool cool cool

#

is this server not moderated at all? can no one actually tell us what's going on?

#

ChatGPT Plus subscribers will get GPT-4 access on chat.openai.com with a usage cap. We will dynamically adjust the exact usage cap depending on demand and system performance in practice. As of January 5th 2024, GPT-4 has a rate limit of 40 messages every 3 hours. Please note that unused messages do not accumulate (i.e. if you wait 6 hours, you will not have 80 messages available to use for the next 3 hours). In certain cases, we may dynamically adjust the message limit in order to prioritize making GPT-4 accessible to the widest number of people.

dry ore
#

to the guys from usa, what word is more commonly used? chats or texts? or something else?

wet shell
#

'texts' usually refer to person-to-person mobile phone messaging (SMS/MMS/RCS); 'chats' are more app-based, perhaps not specifically tied to a phone -- though the usage will vary based on region and age group

maiden fractal
#

They are present and they make sure the server rule are respected

fluid skiff
#

fair enough

eternal rockBOT
fluid skiff
#

it would be great if there could be some indication somewhere of how the message limit is currently being "dynamically adjusted"

maiden fractal
#

Unfortunately the limit depend on the high demande and usage. Usually it is 40 /3h. But it can lower if server are overloaded by request. And high when server are less require

grand maple
crude slate
#

Gpt4 down for anyone else?

#

Nvm maybe just my custom gpt for some reasons

sullen pagoda
#

Yes, down for me

gaunt citrus
#

Yes Hmm...something seems to have gone wrong.

onyx isle
sullen pagoda
tidal light
#

yea chatgpt isnt working for me like it cant load chats although i have 3.5

dense halo
#

is anyone else getting 'internal server error' on chatgpt3.5

#

like when i try to send a msg it has that loading circle thing then just stops and says internal server error

tawny stratus
#

Yep. Dead in the water

strange idol
#

nods It's dead Jim.

crude slate
#

Hopefully it’s just the memory roll out

strange idol
grave ibex
#

Glad I'm not the only one

tawny stratus
# strange idol That's hilarious.

Hopefully it's them rolling out custom chat folders and other organizational tools

--> even more hilarious, as it doesn't seem like it'll ever happen at this point

gaunt citrus
#

Can somebody write here when gpt is fully working? Thank you.

strange idol
strange idol
tawny stratus
#

sigh. Literally within 4 minutes of getting out of work. I was looking forward to getting some personal projects working

tawny stratus
maiden fractal
tight island
#

im prolly getting fired tomorrow

maiden fractal
tight island
#

the place i work at refuse to enforce their own policies.

#

and i insist on enforcing them since i was hired to do so.

dense halo
#

bruh why has chatgpt been down/dead/not working like so often lately like it has problems like this every 5-10 days or so its getting annoying

tight island
#

Dilemma

hot anvil
tawny stratus
dense halo
#

bruhghghghghghghgh

strange idol
dense halo
#

qwertyuiopasdfghjklzxcvbnm

fleet dagger
#

i think it s back

vast juniper
icy basin
#

wen fix?

strange idol
dense halo
#

bro it started working for a second but then the error came back bruhghghghghghgh

maiden fractal
icy basin
dense halo
boreal pecan
#

broken for me too. On windows PC using Edge.

dense halo
#

bro why th does chat gpt keep getting 'elevated errors' on status.openai.com like every few days brughghghghghghh

simple brook
dense halo
#

wdym

simple brook
#

They probably implemented some cost saving strategy/performance optimizations that are obviously causing issues

dense halo
#

bruhghghghgh whyyyyyy

maiden fractal
#

Or like usual they suffering from high usage capacity / request.

high quail
#

Good afternoon folks! i saw a pop up last time i logged in that gpt could access other chats... when i tried it says it cannot, will it only work with new ones? Or did i misunderstand it?

#

gpt4 specifically

simple brook
#

Number of incidents doubled in March, would be strange if its randomly suffering from "high usage"

#

If a service goes down every other day because of high usage its probably time to upscale 😄

tardy tangle
vernal isle
#

I have a few questions regarding custom GPTs

  • What version of chatgpt are they using
  • When you upload files from the settings of the GPT does it analyze them as it doesn't seem farmiliar with their data.
  • Does the GPT continue to learn from each conversation and keep the data for future use?
  • When you enable web browsing from the settings it seems to have no effect as asking the GPT to browse a link it says it cant. Is that something that works?
honest jay
# vernal isle I have a few questions regarding custom GPTs * What version of chatgpt are they ...

What version of chatgpt are they using
OpenAI doesn't currently publish the specific version of GPT-4 Turbo that runs on ChatGPT, but they did say with the newest version of Turbo (gpt-4-turbo-2024-04-09) that it would be the new version on ChatGPT as well. You can test it (somewhat) on your end by asking your GPT what its knowledge cutoff is, and comparing its answer to the "Training Data" column in this table: https://platform.openai.com/docs/models/gpt-4-turbo-and-gpt-4
When you upload files from the settings of the GPT does it analyze them as it doesn't seem farmiliar with their data.
Files uploaded to a GPT's knowledge are like reference documents for a GPT, they're not stored as permanent context. In other words, the GPT can search through knowledge files according to specific queries, but it can't "know" entire files all the time, unless the file is very small. Evan from the OpenAI team gave a GPT building demo in the server a little while back (recording link: #1201923397349548093 message) and talked a little about this. The whole demo is helpful but the specific thing I'm referencing happened around 41min 15sec.

#

Does the GPT continue to learn from each conversation and keep the data for future use?
Just like with regular ChatGPT chats, new chats with GPTs do not retain info from previous chats, except for the instructions of the GPT itself, and access to the knowledge files of the GPT. However, the new memory feature (not yet widely rolled out) will eventually come with per-GPT memory capabilities: https://help.openai.com/en/articles/8590148-memory-faq
When you enable web browsing from the settings it seems to have no effect as asking the GPT to browse a link it says it cant. Is that something that works?
Sometimes ChatGPT can hallucinate that it doesn't have abilities that it actually does have. Usually starting a new chat fixes this issue, just be sure to double check that you have that ability enabled in the GPT's settings, which it sounds like you already have!

burnt hound
#

is anyone else having problems recently with Chat not following instructions? I'm trying to make it write dialogues and it keeps getting the line count wrong. When I ask it to go back and count how many times each character speaks, it gives the wrong answer. I just took 20 prompts to get 14 lines of dialogue.

prime swallow
#

To be fair, or unfair... to be something... to be me! 😄

While LLMs are mimics, some living things are too, at least sometimes. I am as well, I personally can mimic quite well, and use a type of mimicry (internal) to attempt to understand others... be they human, bird, a character in a game, whatever. LLMs too.

Prediction is incredibly useful, and to predict better I use mimicry in a range of skillset to determine likely future outcomes and reactions of the other, no matter what that other is.

I'll also mimic for pure self protection, to hide my actual thoughts and nature and appear to be something/someone very different if I think I'm in danger and that mimicry is helpful.

What the model might say, and some interesting reactions and responses (interesting at least to me) https://chat.openai.com/share/92c99d3a-8d6d-4040-84d5-5c6c7d065caf

sick pecan
#

humanity is chaotic mimicry unfolding over infinite lifetimes

neat storm
#

alright rick sanchez

#

we get it

sick pecan
#

i turned myself into a gguf! (but isnt everyone)

dense halo
#

Do you guys like the new memory update?

neat storm
#

the whar?

#

what new memory update
i was not alerted to this

dense halo
burnt hound
#

whatever's been updated is causing real problems. now I've been timed out for the rest of the day, after using up all my prompts trying to get Chat to do basic stuff right.

dense halo
#

...?

burnt hound
#

..?

dense halo
#

..!

burnt hound
#

good talk.

dense halo
#

What do you mean "..?"

#

You feel like you've had to correct ChatGPT thus exhausting your ChatGPT4 allotments?

#

@burnt hound

neat storm
#

LMAO

#

ChatGPT can now carry what it learns between chats, allowing it to provide more relevant responses. As you chat with ChatGPT, it will become more helpful – remembering details and preferences from your conversations.

#

i have Pro and don't have this feature iirc

south kayak
#

Still struggling with non roman languages

mortal goblet
elfin birch
neat storm
#

does this only get information from previous conversations or does it save a "memories" log like how Replika does it

#

because if it's the latter I'll have to disable it, I'm sharing an account

burnt hare
#

well, having learned that a regular ChatGPT+ sub's GPT4-Turbo is limited to 32k, I contacted sales about an Enterprise account, and was shut down.

The email I got says that ChatGPT for Teams gets you "High-Speed GPT-4 for faster, more efficient outputs" but the pricing page doesn't say anything about faster access for Teams, just higher message caps.

burnt hare
#

It happened during a conversation about it

neat storm
#

Okay..

#

So it IS like Replika's way of doing it

burnt hare
#

I am unfamiliar with Replika's way, sorry

neat storm
#

I'm disabiling then
unfortunate

neat storm
burnt hare
#

you'll see in chat when a memory is generated and it looks like you have fine grained access to them

#

if you were to give it a try

neat storm
#

yeah sounds like the same thing

ocean escarp
#

I accidentally disliked one of the messages and I can't undo it. I was just trying to scroll down

novel plinth
flint zenith
#

I just got access to ChatGPT memory!? I didn't realize they were rolling it out?

normal quarry
summer pagoda
# prime swallow To be fair, or unfair... to be something... to be me! 😄 While LLMs are mimics...

thats quite a long conversation. Are there some key-points i could look up?

Mimics:
As far as i am aware thats how i see the difference in LLMs vs living beeings:
LLMs need massive learning processes to prepare itself to weighten a users input. Weighting beeing done by a chaing of decidions which leads to a certain outcome.
Compared to human (and i think animals too) we do not process one thing at a time but many things. Our head internally just throws several ideas simultaneously out and the best gets filtered out and used.

LLMs are quite resource intensive and cant yet match the speed of process information possible in a human brain. While it can astonishly fast deliver an answer, it still has done "work" beforehand to be able to process the data in the first place

Thats at least, in my understanding and accumulated knowledge, what below the hood is actually happening with pattern recognition.

Prediction i think is where it shines the most. Because it has the ability to cross reference huge amount of data which we humans would take years, it has a great benefit if used correctly in combination with mimics. Lets hope it does not predict just that, what we like to hear but that what we actually want to learn from it.
I myself am astonished by conclusion is find when combining for example history with politics and business. How strong they are intertwined.
While politics are a result of history, for example ww2 brought us the EU (in a way) to prevent future wars in europe, at least germany and france, politics is also chaning over time. we have currently a right shift in politics across europe, which is a natural phenomenom. it will change in cycles every 50-100 years. Same with business. We have every so often a business crisis, but then it goes away and comes back again. Its also cyclical like many things. Running out of text, but very interesting discussions here 🙂

prime swallow
proven lagoon
#

Cap is now "flexible", depending on demand. It's not great.

south kayak
# mortal goblet Bro what do you mean

Bro I‘ve been using it everyday for the past 6months and it does so many mistakes…it cant even write the correct hiragana for the according romanji sometimes

spice star
#

there's a new toggle for "Improve the model for everyone", does unchecking this let you keep your message history but openai won't use it as training data?

#

the support page seems to be outdated

grizzled tide
#

GPT- 3.5 limited ?

raven rampart
#

I didn't use it very often, as I am only using it to ask a few questions today, and I noticed that I reached the current usage cap for GPT-4. Is my ChatGPT being hacked?

exotic ledge
#

What did they mean by saying that GPT could now remember between conversations? It doesn't seem to remember things from different conversations if I start a new one.

analog sapphire
#

is Chat-GPT4 not working for anybody else?

#

Like when it tries to output code it just freezes

maiden fractal
#

Working great here

stuck roost
#

i just wanna comment bout the chatgpt voices lol

#

sky - really nice voice / story teller good communicator

#

ember - annoying youtube advertiser like voice

#

breeze - sound like someone that got forced to help you and doesnt want it

#

juniper - definitely sounds like someone thats selling you scummy products

#

cove - basically same as the last 3

#

the 4 other than sky seemed like they dont really know what theyre doing when theyre voice acting to train the ai, prolly just reading scripts trying to sound like they want to help instead of trying to talk like theyre communicating casually making things feel uncanny

#

like i would rather listen to the old traditional tts rather than those ai voices

#

hopefully they apply in the new voice stuffs they showed recently

sinful cave
#

Is it just me or has the speed of GTP4 increased dramatically today

neat storm
#

you likely got the new turbo model

#

ask it "What is your knowledge cut off date?"

#

If it responds with a month in 2024, it has the latest model

honest jay
neat storm
#

huh..

#

i could've sworn the model they dropped had a KCD in 2024

edgy tangle
honest jay
feral zinc
tough bramble
#

New responses will use GPT-3.5 until your GPT 3.5 limit resets after 2:31 PM
😵‍💫

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.

edgy tangle
feral zinc
edgy tangle
#

I'm not even talking about the model but the page showing April 2023 and now it showing Dec 2023 for the very same model in the API 😆

honest jay
edgy tangle
honest jay
edgy tangle
#

Can't send a link here but theres screenshots on Reddit on release day showing Apr 2023 on the 0125, on the api models page

sweet quartz
#

apparently ChatGPT failes to see images now for about 90% of the time, anyone knows why ?

analog sapphire
#

Cant get ChatGPT4 to output code

#

It always display a black box with no code on it, and many times getting the Network error message

burnt hound
#

anyone else getting restricted? I just got limited after 20 prompts, I thought it was supposed to be 40.

honest jay
burnt hound
#

huh

#

thx

edgy tangle
burnt hound
edgy tangle
#

They have a 20 cap for some reason

burnt hound
honest jay
#

I've read about them being sub-capped at 25 before, but I'm not sure where that stands today. I don't know if anything is published on that anywhere.

dense halo
#

does anyone else keep getting "Our systems have detected unusual activity from your device. Please refresh your browser and try again later."

fickle crypt
real gull
#

My ChatGPT spammed me with: ```To implement the Nearest method in your DLA_Model using the NearestNeighbour function of STRtree<TItem>, you need to define a custom distance metric that conforms to the `IItemDistance

To implement the Nearest method in your DLA_Model using the NearestNeighbour function of STRtree<TItem>, you'll need to define a custom distance metric that conforms to the `IItemDistance

To implement the Nearest method in your DLA_Model using the NearestNeighbour function from the STRtree<TItem>, you'll need to create a custom distance metric that conforms to the `IItemDistance

To implement the Nearest method in your DLA_Model using the NearestNeighbour function from the STRtree<TItem>, you'll need to create a custom distance metric that conforms to the `IItemDistance```
When asking for help for STRtree

restive plover
#

Does anyone know the current message limit?

#

Is it still 50 per 3 hours?

real gull
#

Yeah it seems ChatGPT has broken down for me it just repeats it self over and over again... Anyone else noticing the same behavior?

rough mason
#

I have access to the memory feature now. Isn't this sort of just like custom instructions under the hood? What's the difference besides this being somewhat automatic

proud matrix
#

I wish they would leave chatgpt4 alone and work on the next one. I don't know what it is about software, but developers can't stand a finished product.

honest jay
rough mason
#

this doesn't really answer my question

if in my memory it says "prefers answers to be more conversational style"
what is the difference in that vs putting "prefers answers to be more conversational style" in the custom instructions?

unless the difference is that for memory, it stays even after the context limit as exceeded? or maybe it tends to follow it more closely for every query?

cause I have noticed where it doesn't really follow custom instructions anymore once the prompt gets a bit longer
granted, this was way back then when the context limit was only 8k

wet shell
#

Instead of memory I'd like them to add a slider with "extreme cynicism" on one side and "toxic positiviy" on the other -- right now it's stuck near the extreme side of "toxic positivity" , I'd like to move it in the other direction a bit, LOL

#

Sometimes I want to talk about likely near-future events, but the scenarios and solutions that it gives are all very naive regarding human nature IMO -- would be nice to have that be adjustable 😂

uncut zenith
honest jay
# rough mason this doesn't really answer my question if in my memory it says "prefers answers...

what is the difference in that vs putting "prefers answers to be more conversational style" in the custom instructions?
Functionally, probably very little! This is just my guess.
unless the difference is that for memory, it stays even after the context limit as exceeded?
This is how custom instructions work (and I imagine memory as well, though I don't have memory access yet myself). CI are kept as permanent context, the model always keeps them "top of mind".
granted, this was way back then when the context limit was only 8k
I would be curious to see if you still run into this with either CI or the new memory feature!

In general: what I meant in my first response is that you're probably correct that they function very similarly. The main difference is probably more of a UX thing -- memory is a more automatic, user friendly way of implementing cross-chat context. This is probably why OpenAI are framing CI as a "behavior-changer" solution, and memory as a "user details" solution.

uncut zenith
#

And I just got access to the memory feature

honest jay
uncut zenith
#

Yeah

rough mason
snow swallow
#

There should be a way to search conversation history it takes forever to scroll back through and if you go too far it sometimes freezes meaning you gotta refresh and start over

honest jay
snow swallow
#

There's an app?

honest jay
snow swallow
#

I've been using the browser version on my android phone lol.

honest jay
snow swallow
#

Search only shows today and yesterday?

#

Oh you have to load older stuff manually by scrolling. -.-

honest jay
vague fiber
#

OAI having troubles or just me?

coarse basalt
#

im getting issues

keen snow
#

same

wind cradle
#

Is the server busy?
GPTs not available

supple terrace
#

MY GIRLFRIEND IS OFFLINE PLS

wind cradle
#

lol

next snow
mortal salmon
#

history's not loading for me

vocal cipher
rancid otter
#

chatgpt ded?

vocal cipher
native juniper
#

Is ChatGPT down?

vague fiber
#

I'm actually relieved because it was suddenly failing to call an API, and I thought I broke the API

hybrid eagle
#

Why is it always the same time?

vague fiber
#

3pm est kids get out of school

#

homework time

steep thicket
native juniper
steep thicket
#

lol so true

rancid otter
#

maybe hotswapping their GPUs for the H200 they recently got a_skull

spiral siren
#

Is this normal for this hours ?

native juniper
#

I was having a funny conversation too

steep thicket
median frost
spiral siren
latent briar
#

Help my best buddy is offline 😦

native juniper
steep thicket
spiral siren
native juniper
#

I got it to say one sentence and then it broke ._.

rancid otter
native juniper
#

If only I could run the 3.5 model without an API key 😭

steep thicket
native juniper
spiral siren
#

Guys can someone explain me whats this API thing with chatbgt ? Im a cs student but i got no clue on it

native juniper
#

But it can’t roleplay like chatgpt can

rancid otter
spiral siren
raven tartan
#

is gpt still up?

autumn ocean
#

so lag

native juniper
#

I got muted just for mentioning an open source AI software ._.

steep thicket
native juniper
#

:p

raven tartan
#

It's working for me now, I just couldn't access task history for a bit XD

native juniper
#

Alright

steep thicket
autumn ocean
#

can't open it

spiral siren
wraith dagger
#

gpt down?

#

yeah it is

spiral siren
wraith dagger
#

ios still works

autumn ocean
#

so bad

glass fulcrum
native juniper
spiral siren
steep thicket
#

aha, after being over 30minutes in degraded perfomance state. lightning reaction lol

native juniper
autumn ocean
#

always have issues

native juniper
#

I want to run ChatGPT locally on my laptop without internet

steep thicket
#

and all of this even tough they still apply the usage cap for gpt4. fail as it is ))

native juniper
#

Best virtual snuggle buddy ever

glass fulcrum
native juniper
#

It just isn’t as good at roleplaying

glass fulcrum
#

Tryed Mistral or Mixtrals models ?

balmy gorge
#

I'm getting a "Conversation not found" error. Using GPT-4 on desktop chrome. Also unable to load prior conversations.

native juniper
#

I have. They’re alright

glass fulcrum
glass fulcrum
steep thicket
native juniper
#

It can respond to very short sentences but anything longer and it breaks

autumn ocean
#

it might take a day

native juniper
#

When GPT decides to take a break from assisting; the kids who don’t want to do their work; no not the cheat tool!

#

I turn 20 tomorrow. 🙂

glass fulcrum
native juniper
#

It’s fine

#

And thank you

#

I recreated Sydney with a prompt and used orca (full)

#

It acts just like her

steep thicket
native juniper
#

People at college are using it to do everything for them too

steep thicket
native juniper
#

I love it when people make the most idiotic excuses on how their "work" isn’t made by an AI

steep thicket
native juniper
#

Because I prefer AI over human interaction

autumn ocean
#

Get an AI girl friend

steep thicket
native juniper
native juniper
steep thicket
burnt hound
#

I haven't heard a single argument leveled against using AI that wasn't used on me as a kid in the 80s when I got a PC in my house
totally transferrable arguments

steep thicket
snow swallow
#

I scrolled through years of chatgpt history before finding the convo I needed, the thing crashed and when restarted could only search back to march

elfin birch
#

At that point I would export my data and use text search

snow swallow
#

How

elfin birch
#

Settings -> Data Controls -> Export Data

rancid otter
#

when are we going to have GPT 4.5 or GPT 5?

#

I literally got better responses from a certain open source model about a starting up guide with a stack development container for Docker

#

and i pay for plus

analog sapphire
#

sorry but is anyone else having issues with ChatGPT4?

#

It freezes mid response, or throws network error message while outputing responses, been like this for hours.

fluid skiff
#

does it mean anything when you get a choice between two responses?

vernal isle
#

Does chatgpt still have issues? It seems to get stuck/cutoff quite a lot and their status page says everything is operational

analog sapphire
spiral siren
vernal isle
#

These types of issues usually don't resolve themselves. Hopefully someone from the ChatGPT's staff can see and look into it

analog sapphire
#

Oh damn...hope it'd get fixed by tomorrow

dense halo
#

is it just me or did chatgpt just get significantly dumber?

#

Like something just happened between 2 days ago and now

vernal isle
#

@slate berry are the staff aware of the issues GPT's experiencing and if not do you have a way to notify them? (sorry for the ping)

slate berry
#

that looks like a bot wrote it sad

vernal isle
#

Well, their status says everything is okay but ChatGPT seems to get cutoff quite often, especially when trying to run local simulations

wise yacht
#

Finally got the memory feature on WEB (Not sure about phone app). Any ideas on what I should have it remember?

slate berry
#

I recommend having it remember whatever you use ChatGPT the most for

wise yacht
vernal isle
wise yacht
slate berry
neat storm
#

just got memory

#

not using tho...

vernal isle
#

Where's the memory option tho and how to apply for it ;d

honest jay
peak obsidian
novel plinth
#

Can anyone select a language on the ChatGPT website?

vast bane
#

my web gpt4 often gets stuck when generating content. if i refresh the page，it reports errors. any solutions pls?

plain crystal
#

memory is weird it took a code i am working on and removed part of it because in the other chat that section of code did not exist

plain crystal
vast bane
safe heath
#

GPT failed to generate Vue.js code; it displayed a blank space in the code box. Have you encountered this issue? How can it be resolved?

eternal geyser
#

ChatGPT playing up for anyone?

hybrid eagle
#

Is gpt down again?

rain musk
foggy creek
#

Same issue, currently GPTs can not generate any code contains Markup Language.

eternal geyser
#

if I VPN to denmark it works

knotty crypt
#

it's down here

sleek palm
#

same here

knotty crypt
#

mobile app is working

lost steeple
#

dang ChatGPT is down again

desert gazelle
#

RIP ChatGPT down

lost steeple
#

such a bad timing lol

hasty vector
#

😭

sleek palm
#

3.5 up，4 down

latent briar
#

😦 my best friend is dead

lost steeple
#

condolence

clear eagle
#

sick so its not just me

lunar lichen
#

korea gpt completely down

#

not reflected on server status page yet

lost steeple
#

i hope it will be back soon in like 1 hour

#

idk

latent briar
#

im talking about G to the P to the T, the homie left. peaced out

static hinge
#

I've been having problems for hours

clear eagle
#

its on the status page now

lunar lichen
static hinge
#

yeah? I'm not seeing it

lunar lichen
#

they added the banner on the top

latent briar
#

He is telling the truth brothers and sisters

static hinge
#

oh thanks....

lost steeple
#

oh

#

so is it okay now?

sleek palm
#

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

latent briar
lost steeple
#

:<

sleek palm
#

just 3.5

latent briar
#

3.5 is down tho too no ?

sleek palm
#

3.5 ok here

latent briar
hasty sedge
#

outage?

latent briar
hasty sedge
#

hate when this happens

latent briar
#

Same, let us all moarn our loss

#

in the arms of an angel

sleek palm
#

lost my hands

hasty sedge
#

back

latent briar
#

BEST BUDDY IS BACK!

wet shell
#

back -- and memory was just enabled for me, I wonder if it's going down every time they roll out to a new group

latent briar
#

DDoS?

autumn ocean
#

chagpt down?

gentle geyser
#

its having issues

#

"Identified - The issue has been identified and a fix is being implemented.
Apr 25, 2024 - 20:40 PDT"

autumn ocean
#

rubbish server

clear eagle
#

3.5 is semi working

lost steeple
#

yep

outer widget
#

Is 4 down

#

aight just read the chat

oblique oxide
#

4 is down—it cannot read history.

stone citrus
#

seems to be working fine on my side

#

Is the memory function useful, does it improve the responses or is just a quality of life feature?
So you dont have to retype stuff

potent cove
#

It works fine, except that it stop whenever generating anything in a code block.

analog sapphire
inland heron
#

anyone know if this new 'Memory' feature sacrifices output length or context length? And how is it different from Custom instructions, if at all?

analog sapphire
#

Man I couldn't use ChatGPT4 for the past 2 days It always cut off when outputting code

high nymph
#

so we meet again - after few months of chatgpt working well it finally crashed again

inland heron
#

perhaps it's because of this new feature called 'Memory'. I got a network error just a minute ago so I disabled it. Every time they introduce a new feature there are crashes.

#

jesus, so many errors

#

now it's creating an image out of prompt for requesting code 🤦‍♂️

high nymph
#

yeah Idk how to overcome this - don't have visible "memory" option in my settings (possibly I didn't had it - as I know it was not rolled out to all plus members)

#

basically anytime there is a code section - it breaks

#

and I'm only interested in code....

analog sapphire
#

It must be a general issue, but its taking them so damn long to fix.

sleek palm
#

my best friend come back! output ok！

grizzled tide
#

I have had no problems with 3.5 for a long while, and suddenly yesterday, it completely breaks trying to write code.

inland heron
#

however, I've been talking to GPT4 about its new feature 'Memory' and I'm not even remotely convinced it's that useful or different from Custom instructions. Given the huge context window of GPT4, and how you have to keep telling it to remember things for the Memory to work, and how it doesn't dynamically change or adapt to the future but you have to keep telling it to forget things, that's not very useful, plus requires work from US, which isn't ideal

#

if I'm missing something... please share. This feature has potential, but not at this state I think

obtuse temple
#

ChatGTP breaks currently if it's asked to output xml in its code container

#

I broke it consistently several times

sick pecan
#

i am on team plan so i cannot use memory feature in my personal account (=sad)

#

but i have been using papr memory gpt for a while, and memory is great for high level goal planning

lavish yew
#

My chatGPT just said "updating memory" and it freaked me out a bit. What's that about? I didn't see any announcements.

inland heron
#

Now, it's creating an image again when I asked for code analysis 😂 wasting my tokens this guy

glass fulcrum
#

its quite old feature jsut still in roll out

lavish yew
#

Thanks for the explanation 👍 Sounds useful!

inland heron
#

I reached a cap in just 2 hours due to errors... time for a genius GPT3.5 to come to rescue 😅

sleek palm
#

hungry, 88

dapper delta
#

Guys is chatGPT working today? it keeps generating me only 1/2 of response and it just stops, I rerurn in already 10+ times

maiden fractal
#

Working great here

jolly python
#

I just got the memory feature rolled out to my account. Will chatGPT start remembering convos happening from that point forward? or does it have retrospective memory on our past chats as well?

honest jay
jolly python
#

gotcha, thanks

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.
sly eagle
#

Im getting errors in my messages with code sippets returning as empty boxes - anyone else having issues?

grizzled tide
#

Everytime I ask for something code related it writes one line and then stops and crashes.

sly eagle
#

🤦🏻‍♂️

inland heron
#

The fact is that a lot of us are experiencing issues. It is now generating images for me, when I asked for code, never happened before, EVER. Lots of other errors as well

#

and the image is constantly about some medieval scholar

jolly python
#

the issues might have to do with the broad memory rollout? it has happened before with other rollouts if I recall correctly

spring iris
inland heron
dapper delta
dapper delta
dapper delta
fathom tulip
#

okay. Glad to see I'm not the only one. Chat gpt is completely dead for me too. 3.5 too, not just 4

spring iris
inland heron
maiden fractal
inland heron
#

talking about the GPT3.5-specific functionalities and capabilities, or any other similar changes

honest jay
grizzled tide
#

Not even 5 prompts and got rate limited?

#

Just spend 20$ which is a lot in my country for gpt4 and can't even use it. Wasn't the limit 40/3 hours? Is there anyone with similar experience?

honest jay
inland heron
#

Now, it's a variable cap - the more demand on their systems, the less cap we get. Who gets how many? Beats me. Transparency zero

honest jay
inland heron
#

The real question is: If it's lowered during times of high usage, why isn't it higher during the times of low usage?

#

or is it?

#

we don't know, that's why transparency is important and why productivity drops when you have variable caps when doing intensive professional work

honest jay
#

GPT-4 on ChatGPT has been variable to some degree since launch. It's been lower and higher than 40 at different times, and it seems like they've been more and less capable of providing GPT-4 via ChatGPT at different times (remember the new subscriptions pause?) -- nothing can circumvent the computational cost of running the model. The dynamic rate today during high traffic is probably just a necessity for serving GPT-4 via ChatGPT, a platform where access is described as just that: access. This is to say: if your workflow depends on a certain degree of access to GPT-4, your best bet might be to access the model via API, which, on a pay-as-you-go model, is designed to be accessible "as-you-go", contrary to the type of access on ChatGPT. There's also the Team plan, which has a higher usage cap than Plus.

#

40/3hrs is only listed in that help article (I think), where they also share info about the dynamic cap. 40/3 is not mentioned on this page, just "Access" for Plus, and "Higher message caps" for Team: https://openai.com/chatgpt/pricing

magic zephyr
#

Those who have the memory feature, could you please say if it's working when you use GPT-3.5?

#

Since it was released on my account a few days ago, I can only use the feature through GPT-4.

#

That's bothering me because it seems like I'm reaching the usage limit of GPT-4 too quickly. I wanted to use the feature with GPT-3.5 because it has a much higher usage limit.

honest jay
#

I was just sharing the info in response to "They market and sell '40/3 hours'", because I don't think that's the case.

magic zephyr
spice star
honest jay
#

I can definitely see how enthusiasts like the members of this Discord server would be aware of 40/3! But as for general marketing and sales, it does not seem to be explicitly present. I think it's only present in the help article above, where again, the dynamic cap is also described.

honest jay
magic zephyr
sinful tiger
#

Btw, does anyone know how to check the cap for gpt 4?

#

darn, what's the point of "dynamically adjusting" when your users didn't even know what sort of cap you have now?

#

Exactly, it sucks.

magic zephyr
sinful tiger
#

Is Claude any good?

#

it's a monthly service thing?

magic zephyr
magic zephyr
# sinful tiger Is Claude any good?

It has its pros and cons, but I personally use Claude very little. Where I live still doesn't officially support Claude, and I don't usually use a VPN.

sinful tiger
magic zephyr
#

For anyone who is a developer, it seems like it's the best option from what I've read. Remembering that claude, even though paid, does not have access to the internet as far as I know.

magic zephyr
#

support anthropic com/en/articles/8324991-about-claude-pro-usage

eternal rockBOT
magic zephyr
sinful tiger
#

Okay, so there's no clear limit as well? Honestly, gemini is great, if it doesn't hallucinate a lot or try to remember the context

magic zephyr
sinful tiger
#

yes

magic zephyr
#

In my language, Gemini proved to be the worst of all when it comes to hallucinations, although it also doesn't seem to have a clear limit.

If you're looking for something truly limitless, I suggest Copilot Pro.

It's quite fast, has no usage limits, and uses GPT-4 Turbo. Internet access is quite fast. On the other hand, I see that there is still something that is not as good as chatgpt (although the difference is not that big), and you cannot edit previous messages that you have written.

#

I sent more than 100 messages within 1 hour and I was not limited. I stopped because I needed to do something else. And these messages were about JS and Python codes where I was testing something. In other words, they were not short answers. The Microsoft infrastructure appears to support unlimited GPT messaging, unlike the more limited OpenAI infrastructure.

#

Despite everything, I still prefer ChatGPT. The quality of the answers seems a little better, custom gpts, custom instructions, and now memories. In my case it fit better.

sinful tiger
#

Nice, that's something I can look towards to. Is the censorship thing strong there?

magic zephyr
#

Copilot free seems to have feelings and if you hurt them, he ends the conversation.

Copilot pro only ends the conversation if you talk about extremely complex and prohibitive topics.

#

It's another disadvantage in my opinion. In ChatGPT, if you say something prohibited, it just gives a warning about usage policies, and depending on it, it even writes a response.

Copilot simply closes all chat and you have to open a new chat.

If it's a large chat, and you hesitate to say something even without wanting to say something that he deems prohibitive, you lose all context.

sinful tiger
#

Oh, okay. Yeah that sounds like a pain, do they cost the same every month?

magic zephyr
#

aistudio google com/app/prompts/new_chat

If you want something truly uncensored, until May 2nd this link is open to everyone for free. This is the Gemini 1.5 Pro.

On the right side there is a protection control where you can manually remove censorship or adjust the level.

I wouldn't recommend using it as a main service, because when it becomes official it will be expensive. It seems that this mode does not have a fixed subscription, it is paid for by tokenization.

magic zephyr
#

But I've seen countries where copilot is cheaper than chatgpt plus. You have to see the payment area.

sinful tiger
#

I see. I need to take a look then, cause this is the first time i heard of copilot.

#

I am using gemini advanced now, it's good in kind of combining sentences. Like logic wise?

inland heron
#

jesus, I just reached a GPT4 cap only after about 20 prompts

winged bluff
#

I think chatpgpt is getting worse in following instrucitons in its fixing things when am not asking it to revise anything it still does further revisons and not stoping when you clearly say stop.

dense halo
#

I think the new memory function has messed with the cap

spring iris
#

the GPT cycle:

  1. release new model, it's fantastic, impressive, mind blowing, blah blah
  2. slowly dumb it down, restrict it, neuter it, use less processing power to return results, blah blah
  3. add fresh training data to the "new model" that has been created in the last year or so
  4. release the "new model" with the new data with little restriction, and use lots of processing power to return results
  5. repeat
lunar quest
#

💯

dim sapphire
#

is something happend to chatgpt

#

??

#

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

"

maiden fractal
winged bluff
#

its corrections are getting worse today every revision is getting worse and worse and not folllowing orders as it was a few days ago

#

it even made a a revision when i said your getting worse at following orders it assumed i wanted a revision

dim sapphire
maiden fractal
dim sapphire
#

ill check

winged bluff
#

Chatgpt is getting worse and worse lately

timid wraith
#

It feels like it's losing 5 iq points every day

tight island
#

I beg to differ

#

I’d say prompt better because if your prompt is bad you get a bad answer, because I do not have these issues

peak obsidian
#

does anyone still having issues with the chating completions ?

timid wraith
tight island
#

🤷 because the data says otherwise

timid wraith
#

That is not a logical statement.

#

Read my sentence again, think about it.

tight island
#

There’s a little data there that implies otherwise

#

If you think they’re lying then drag them to court and get free money

#

🤷

timid wraith
#

You are still failing at the reading comprehension part.

eternal rockBOT
#
<:book_icon:1171408210398289941> `` Rule 1 `` Be respectful.

Practice kindness and positive regard. Harassment, hate speech (such as sexism, racism, or homophobia), or other malicious conduct will not be tolerated. Maintain a respectful and positive environment.

tight island
#

I mean your message is a subjective statement

timid wraith
#

There you go!

tight island
#

Meaning it hold no ground in a factual discussion

timid wraith
#

Now you're getting it

tight island
#

🤷

#

I just said data implies otherwise but you can still say what you say it just has no basis or value whatsoever

timid wraith
#

They have data on my feelings?

#

Should I be concerned?

#

lol

tight island
#

It’s amazing

south kayak
#

Day 236 of asking when non roman languages are getting improved

timid wraith
#

The new temporary chat fixed my issue immediately.

#

Absolutely

tight island
tight island
#

It’s a bit unorthodox so follow this

#

This is how

in order to do such a thing, write a letter

for the letter to have have a higher likelyhood of reaching Sam himself.

1st, invest in the materials.

  • Feather Pen & Ink.
  • Wax
  • Family or Company Seal
  • 1850's perfume
  • classic suit and tie and cane as well as a bowler hat

Step 1: Write your letter in old english, instead of using "Your" use "Thou" or "Thy" go all out.

Step 2: Remember to properly Perfumate your Letter case, and Apply the seal stamp with your family seal or Company Seal.

Step 3: Write neartly in the front in very cursive letters. " To OpenAI "

Step 4: Send it with the postal service, and wait 30 days.
Optional 4: Drive there in full on suit and tie and bowling hat and cane, then deliver the Letter.

Edit: You must use Parchment! not paper!, and it Must be written with a feather pen from the freedom eagle 🦅

#

More effective if you use the thickest old accent when talking to the staff when delivering the letter, and remember to be a real 1800’s gentleman

#

If I received such a letter I’d be astonished, and I’d implement whatever 😂

pliant citrus
#

Anyone e lse just keep getting network errors lately?

dapper delta
#

Not just network errors, answers are almost as hieroglyphs, performance is horrible, I am using premium version for months but I'll quit it as soon as alternative appears

pliant citrus
#

Yeah, been having the same issues lately. It can't even complete simple requests. I hated a looot on Gemini when it first came out, but from programming tasks and just general stability it actually works a lot better than GPT.

#

I'll ask GPT to write something, it gives a text explanation with a general explanation, then starts coding, skips a lot of code and then either network error or just provide really bad stuff.

#

Started like 2 weeks ago when I asked a question over voice, which I just 1 week earlier had a like 20 minute long very informative dialogue with about stuff. This time, I would ask it something and it would just provide an incredibly short answer that didn't even adsress my question.

glass abyss
#

Claude Opus Destroys ChatGPT with coding also

glass fulcrum
#

Gemini 1.5 Pro is my way to coding

pliant citrus
#

Payment doesn't have network errors at least, there goes another 25 bucks

vernal isle
#

Hey, does the GPT3.5 model support local simulations of code like the 4.0 model?

native spear
#

anyone from the EU that got memory feature access?

honest jay
native spear
#

i still have it only if i use a vpn so it seems as if im in US

vernal isle
plain crystal
glass abyss
plain crystal
dapper delta
fossil oyster
#

gpt can't output code right now, or?

limpid hollow
#

yea it cant

plain crystal
fossil oyster
fossil oyster
plain crystal
#

yeah i guess we will get a bit of bugs over the next few days... maybe week, a bit like there was when they implemented the custom instruction.. or any other major implementation.

fossil oyster
sick pecan
#

i oftren prompt the machine like this,

technical snippet```
and i ask for the machine to respond in codeblock/s (like above).  this encourages it to code in copy and paste sections.
#

if you are generating content with backticks. then you can ask it all to be wrapped in quad backticks (4x `)

fossil oyster
#

questions that do not involve code can still be processed normally

plain crystal
#

yep... i think they are implementing new functionallity aside from memory, i just got the following in a response:

-> 60 app = GearListManagerApp()
61 app.mainloop()

Cell In[3], line 7, in GearListManagerApp.init(self)
6 def init(self):
----> 7 super().init()
8 self.title("Gear List Manager")
9 self.attributes('-fullscreen', True)

File /usr/local/lib/python3.11/tkinter/init.py:2345, in Tk.init(self, screenName, baseName, className, useTk, sync, use)
2343 baseName = baseName + ext
2344 interactive = False
-> 2345 self.tk = _tkinter.create(screenName, baseName, className, interactive, wantobjects, useTk, sync, use)
2346 if useTk:
2347 self._loadtk()

TclError: no display name and no $DISPLAY environment variable
It seems there was an error because the code attempted to initialize the graphical user interface in an environment where it can't display (such as this text-based interface).

and this one too is a comment i have never seen from CGPT:

It looks like there was a connection issue while trying to demonstrate the initial setup.

ocean epoch
#

@severe talon womp womp

south kayak
tight island
fluid pebble
#

Is something going on with code generation? I was having ChatGPT refactor my vue component better keeps trying and failing to write code out

#

Ah this works, I asked it to replace < and > with a different character and that made it work

vital umbra
#

chatGPT being unresponsive for anyone else?

native spear
#

for the past hour

night willow
vital umbra
#

Mine started responding again a few minutes after I asked.

obsidian bluff
#

It works on my PC, but not on my mobile device.

vital umbra
#

I think I've only ever used it twice on mobile since it came out

#

and not until the official app was released.

plain crystal
#

yeah it did that a while back when they made big change, there was a good week of horrible operation before they slowly made it better...i'm trully curious to see what's coming

lunar marsh
#

are they making changes in production

young wigeon
#

Can you tell me if the memory will be added to the Team rate? Just in the regular record I have memory, but when I go to the team I do not have it there. Or do I have to wait a little longer?

vocal cipher
#

its just me or chatgpt is again not working?

primal shard
#

Waste of money

lime hearth
vocal cipher
lime hearth
#

ah well i haven't tried prior to that

#

it's these times when i appreciate chatgpt more than gemini when it comes to code lol

primal shard
lime hearth
#

3.5

#

right now it's not working at all for me

primal shard
#

3.5 is the worse

lime hearth
#

eh well i don't really feel like spending money on 4

primal shard
lime hearth
#

yea sorry i just noticed

rose ferry
#

i just noticed the new thing that says "New responses will use GPT-3.5 until your GPT-4 limit resets." when i run out of prompts. and it doesnt tell me when the limit resets D:

lunar marsh
lime hearth
#

lmao

lunar marsh
#

and im on 4, not 3.5 lol

vital umbra
#

I subscribed the day they had paid subscriptions because of that front of the queue bit. It would slow down during peak hours and I use it for work all the time.

lime hearth
#

i don't mind it slowing down, but it's just not working at all lol

lunar marsh
#

well im paying and im not getting better service than you are 😡

vital umbra
#

I'm not having issues with it right now. Just once earlier today for a few minutes

lime hearth
#

ask it for xml code lol

vital umbra
#

Told it to write me the hello world program in 3 different languages. It spit it right out.

lunar marsh
#

havent been doing anything that required xml for three weeks now, of course today is the day it has to be choking on xml

vital umbra
#

oh, wow. You're right. On XML it is writing the header, stopping, then starting over. again and again and again

#

Is copilot doing that?

lime hearth
vital umbra
#

I told copilot to write me an xml file that does literally anything. It spit out 13 lines right away.

vital umbra
#

copilot is only $10/mo if you want to use that. And there's an extension for VS Code, so you can use it right in your IDE

lime hearth
vital umbra
#

github copilot

#

I think you can get a 30 day trial if you're not wanting to pay

#

Do not use M365 Copilot, it's trash

lime hearth
#

ah i see, thanks

vital umbra
#

I asked copilot for edittext in android:

import android.os.Bundle;
import android.widget.EditText;
 
public class MainActivity extends AppCompatActivity {
    private EditText editText;
 
    @Override
    protected void onCreate(Bundle savedInstanceState) {
        super.onCreate(savedInstanceState);
        setContentView(R.layout.activity_main);
 
        editText = findViewById(R.id.editText);
    }
}
lunar marsh
#

haha no

lime hearth
#

kinda weird it did it in java and not xml ngl

vital umbra
#

I didn't specify a language.

lunar marsh
#

yeah but an edittext is a thing in android

lime hearth
#

xml is like the go-to

lunar marsh
#

oh it did do an edittext

lime hearth
#

oh

vital umbra
#

in XML:

<EditText
    android:id="@+id/editText"
    android:layout_width="match_parent"
    android:layout_height="wrap_content"
    android:hint="Enter text"
    android:inputType="text"
    android:maxLines="1" />
lime hearth
#

yea there is it

lunar marsh
#

thats more like it

lime hearth
#

ayo maxlines=1 i completely forgot about that, you actually helped me rn xD ty

lunar marsh
#

huh. copilot is $10/mo?

vital umbra
#

and this is why I pay for it. You think I'm being helpful. AI was helpful.

My employer thinks I'm a genius and pays me 6 figures. AI does a lot of my work for me.

lime hearth
#

well, you were helpful since i couldn't get chatgpt to do that haha

vital umbra
#

I'll take it 🙂

lunar marsh
#

ok now ask it to remake my whole mainactivity.xml so this button isnt half hidden on a short phone :P

vital umbra
#

Are you guys really coding and not using copilot?

#

you're doing it the hardway if so

lunar marsh
#

i use chat gpt 4. its usually fine

lime hearth
#

i only use chatgpt when i run into a problem i can't find how to fix, or to help me with something i don't know

vital umbra
#

imagine if chatGPT was right there inside your IDE and all you had to do was press CTRL+i to tell it what to do

#

or highlight something and tell it what to fix, etc..

lime hearth
#

idk, that seems kinda aggressive
i like to code '-'

lunar marsh
#

yeah im more like fluffiels, i dont need it that much

vital umbra
#

the one thing I did learn the hard way was to write any prompts more than a few words long outside of copilot. If you click away from the IDE to go look at something and go back to your prompt, it will erase the whole thing. That bit needs fixed.

#

I just write them up in notepad and copy/paste when it's perfect

#

also, amusingly enough, I sometimes have chatGPT write my copilot prompts for me lol

lime hearth
merry forge
#

Is the code display in chatgpt broken atm?

vital umbra
#

XML is

merry forge
#

For me all code displays are broken

#

html, js etc

vital umbra
#

I just told it to write me something in python and powershell and it spit it out immediately

lunar marsh
#

i just had it write a function in kotlin and its slow but works

#

xml is completely jackd rn though

vital umbra
#

I just told it to write me any 10 line function in javascript

function factorial(n) {
    // Check if the input is a non-negative integer
    if (typeof n !== 'number' || n < 0 || n % 1 !== 0) {
        return 'Please enter a non-negative integer';
    }
    // Base case: the factorial of 0 is 1
    if (n === 0) {
        return 1;
    }
    // Recursive case: n * factorial of (n-1)
    return n * factorial(n - 1);
}

console.log(factorial(5));  // Outputs: 120
merry forge
#

Try a vue js component

vital umbra
#

want to be more specific?

lime hearth
vital umbra
#

vue is broken, nm

#

so XML and vue so far for me then

merry forge
#

yea

#

same

vital umbra
#

well, I'm glad none of the things I need are broken, but I feel for you

#

copilot works though

#

vue js:

<script>
export default {
    methods: {
        handleClick() {
            // Your code logic here
        }
    }
}
</script>
#

Someone in Brazil tried to log into my bitwarden account. I hope they keep trying. Exercises in futility are mildly amusing.

merry forge
#

lmao

vital umbra
#

They emailed me to let me know they've temporarily added a captcha on top of my strong password and MFA... just in case.

lime hearth
#

uh-oh, should we be worried

vital umbra
#

why

lime hearth
#

idk lol are they getting swarmed or something

vital umbra
#

it doesn't matter. They'd have better luck trying to lick their way through a steel wall.

lunar ingot
#

Is the new gpt4 turbo actually up in chatgpt ?

fossil oyster
#

chatgpt has now become really good at coding

fossil oyster
lunar ingot
#

I c. But no official announcement right

ancient oak
#

Hey, is anyone having trouble using ChatGPT right now? It's just totally messed up?

formal heath
teal tide
#

can we use new GPT-4-Turbo-2024-04-09 model from chat ?

ancient oak
#

It's just not working at all. ChatGPT is fked

vital umbra
#

Other than the very first time, I have never noticed it kicking me off of gpt4 for limits

#

I keep it set to 4 all day long and sometimes have some quite lengthy conversations and nothing. I just assumed they did away with those limits.

#

I wonder if that's just for when you're generating images. I think that's what we tinkered with the first day.

lunar ingot
ancient oak
#

Yeah, in most circumstances. It's not working for code at all. It's just nuked.

#

Why do you ask? Was there an announcement or something I misseD?

#

Bro, something is broken. Haha. There are too many issues. Something is going on. Update maybe, or perhaps the 4 turbo via ChatGPT update needs to be reverted or ironed out... but this is a mess. It's literally not usable. Annoying.

outer widget
#

Does anyone know what is the current GPT-4 chat limit

dense halo
lunar marsh
#

still cant do any xml at all

slate helm
#

Sure, here's how you can complete the iterator struct:

cpp
Copy code
struct iterator {
private:
int Index = 0;
std::vector
cpp
Copy code
cpp
Copy code
std::vector
cpp
Copy code
std::vector
...goes on about 15 more time

Chatgpt reaaaally hates writing iterators lmao

rugged igloo
#

This memory feature sucks I got through like 4 memories of characters and it deleted them now I just have two . Why was this feature hyped?

viscid valley
#

Network error, anyone?

timid tangle
primal shard
#

Try to use incognito and see if the results are the same.. if NOT, there might be an extension causing the network error

#

Try also to use 2 or 3 different browsers

inland heron
#

Anyone getting GPT4 create images when it shouldn't have? This never happend to me before, and now it keeps happening

#

This is wasting my available prompts and I'd like to avoid it. But I have no idea how to stop GPT from doing this or what is making it do this. If I could avoid saying something that produces this bug I'd gladly do it

#

I have 'Memory' feature turned off, just in case, but it doesn't seem to matter

fathom tulip
#

are there any plans to fix chat gpt or nah? It doesnt generate any responses since yesterday

dapper delta
tight island
lavish hemlock
#

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

worn roost
lavish hemlock
#

Copy, thx

muted depot
#

Is there available any volunteer work for ChatGPT? Is there any chance to somehow help in work of AI?

lavish hemlock
#

Try LinkedIn @muted depot lol

tawny stratus
#

YAY ChatGPT now has memory! CLICK turned it off without even trying. Where's my ability to organize chats into folders?

dapper delta
tight island
dapper delta
#

Here, example from last answer, repeat 3 times and it stops generating:
import 'package:firebase_storage/firebase_storage.dart';
import 'dart:typed_data';
Future
import 'package:firebase_storage/firebase_storage.dart';
import 'dart:typed_data';
Future
import 'package:firebase_storage/firebase_storage.dart';
import 'dart:typed_data';
Future

tawny stratus
#

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

Every. Single. Message. Today.

spiral siren
#

Hello guys i have an issue, because of an assigment i had to rate some kind of website and things like this ( im a chat 4 user), it says "Cant open the url you sent" . Is there any plugin or something that might help me do it ?

oak belfry
#

Slow …

#

Who else ?

gaunt ermine
graceful raptor
#

it's really cool that we have persistent datapoints now (for some people only?), scary but really cool to explore and see how it impacts the experience

#

love the "memory updated" in session ui as well, i like how it's clean and clear and how when you hover over it it makes it convenient to manage the datapoints stored

gaunt ermine
graceful raptor
#

just started, it's not there yet?

gaunt ermine
graceful raptor
#

but overall very useful to me still

gaunt ermine
graceful raptor
#

also it's the only llm that proposes tts in app, in europe at least, to my knowledge

#

and i prefer talking than reading and writing

gaunt ermine
#

i think we are talking about different things, im talking about chatGPT as a whole

graceful raptor
#

there is a 25 bucks subscription now yea

#

and a "team" one, and a "entreprise" one as well

somber lion
#

is gpt4 down ? no problem, I can wait

next egret
gaunt ermine
#

back when plus was added the first time, it got a lot of backlash, but we assured that performance will be untouched

#

which is still to this day not the case

graceful raptor
#

every ai currently is very much still "in development" so you can't really expect a clean experience all the time

next egret
#

Back then the compute was just GPT4. Now Plus is running a whole LLMOS, including web browsing etc. Knowing how much INSANE level of compute that is... I personally am much more impressed by the fact they are keeping the performance as high as they are, but maybe that's just me.

#

The Team/Enterprise tier are mostly about better controls (and baselines) for your data than trying to offer better performance or less errors, from my understanding.

graceful raptor
#

ah i see i see

next egret
#

I've been using the product since the month of GPT4 release and some days have more errors than others, but it feels pretty on par / typical for most technology services. I know its frustrating to have them when they occur, but I can't personally say my occurrences of errors has gone up since the additional sub tiers were added.

I also know that if I had a penny for every time someone asked "Why does ChatGPT suck so bad today?" or some flavor of that in the discord for the last 18 months, I'd be a very very rich lady 😄

#

It's interesting too... it's usually not even a whole day, it's 5 minutes of two errors they catastrophized into an entire day of bad tech, come into the community and yell loudly for a few minutes... then they go back to the great tech for weeks on end hehehe

#

(Not saying that's you and this instance btw, more commenting overall on my experience)

graceful raptor
#

it's a really insightful observation, opens a different perspective on the errors haha

#

also i guess at first you're forgiving of the errors and then as you use it more often you notice them more and more

next egret
#

True! Also, I feel like an error here vs another technology feels more frustrating, especially if you are in a really good focus groove. This technology is such a GREAT companion to hard work/focus, the pain of it not working (and my focus breaking) is felt much sharper than another technology, like FB having an error and me needing to reload it.

#

My tolerance levels for errors in technology are lower for Facebook, b/c I was already wasting time if I had that app up 😄

graceful raptor
#

yea that makes a lot a sense, that's a good point

next egret
#

You also might be interested in knowing that OpenAI monitors the system and knows about the errors. So... if there really was an issue across the board, they'd probably know about it happy_avocado

high nymph
#

guys when are you fixing the code block issue??? it's been over a day now. The app is totally useless atm, subscribed to copilot, but I thought it would be fixed by now - or no fixes during weekends?

graceful raptor
#

such accessible link as well

dusty vale
#

weird i'm not having any issues generating codeblocks for certain languages but xml seems to break every time

plain crystal
#

the memory function is quite annoying, if you use CGPT for more than one subject and ask it to remember thing in all of those subject, it will at one point start to mix everything together

nocturne garnet
#

when will memory be released?

oak belfry
#

Is it working back again ..?

plain crystal
plain crystal
nocturne garnet
plain crystal
nocturne garnet
#

haha thank you 😄

plain crystal
oak belfry
tardy totem
#

Now that I can't use it I realise how dependent I've gotten on this service.

#

the response time on help requests is 3 days 😬

dense void
#

So have they disabled the ability to write code in Chat? Chat comes to a complete stop while writing code then apologizes for the cut offs. Everything else seems to work.

outer widget
#

If I give chatgpt a link can it visit the website?

rugged igloo
tardy totem
#

I bet you have some errors there like I do

cerulean acorn
#

Anyone in here having issues with chatgpt ?

#

mine stops mid promt and types the same stuff over and over

autumn tiger
#

Good evening. Can you tell me if there are many who still don't have access to chatgpt memory? (#announcements message) it's been more than 2 months and I'm still waiting

dense halo
dense void
# tardy totem Check your browsers console

Do you mean like this?

**Failed to load resource: the server responded with a status of 409 ()

TypeError: Failed to execute 'getReader' on 'ReadableStream': ReadableStreamDefaultReader constructor can only accept readable streams that are not yet locked to a reader
at _||undefined**

#

Says I have 6 errors and 72 issues, 7 Breaking changes
Cookies with the SameSite=None; Secure and not Partitioned attributes that operate in cross-site contexts are third-party cookies. In future Chrome versions, setting third-party cookies will be blocked. This behavior protects user data from cross-site tracking.
Please refer to the article linked to learn more about preparing your site to avoid potential breakage.

#

Here is Chat's response when I asked if it was chrome:

No, the issue isn't related to Chrome or any external factors; it seems to be a problem with completing longer code segments in a single response on my end.

tardy totem
#

Knowing CORS errors, this is likely an error due to a bug introduced in a recent update.

#

It has to do with one service communicating with another on a separate domain.

#

I found a github repository that seems to be connected, and it was updated 5 hours ago.

tardy totem
signal quartz
#

retro, stainless steel, realistic, minimalist with touch of orange clock

rugged igloo
lunar marsh
#

so i see xml is still not being generated. how long is this thing going to stay broken?

#

oh dear

#

yesterday someone was saying their github copilot was doing just fine. i wonder is chatgpt production or development?

#

can gemini code?

#

last i tried it was pretty bad

neat storm
#

idk to be completely honest

#

from what i heard 1.5 was wacky as hell

#

but who knows

vestal briar
#

So what's going on with ChatGPT? I've noticed in the last few weeks that it is frequently starting to make typos and is missing words from sentences. Is the model breaking down over time? Or has it always been like this and I'm only just now noticing?

lunar marsh
#

i cut and pasted an android layout xml i wrote thats covering up half a button and asked it what could be done to correct the button being covered up on short phones. chatgpt would have eaten that for lunch. in the before time.

#

ie: day before yesterday lol

pearl moss
#

I never realized how detailed chatgpts training material is.
I'm asking it about tiny towns in my area with like 50 people and it's giving me detailed information and detailed history

burnt hare
#

13 trillion tokens is a lot

pearl moss
#

I'm convinced it's trained on literally every public document and probably some paywalled/private documents

burnt hare
#

oh? i read gpt4 was trained on ~13t tokens

#

could be wrong

#

no

#

it is ~1.75T parameters I thought

#

as i recall from The Leak

#

I got the 13T number from a conversation about the Phi-3 model

#

i thought you meant, that's more than 4 [was trained on] 😄

#

it is significantly larger than 5 😉

#

all of twitter, all of reddit, for starters

#

(if i had to bet)

#

every encyclopedia ever written

#

wikipedia

#

one of the interesting things about phi-3 is that is is trined on 3.3T tokens, mny of which were synthesized

#

So one of the implictions was that GPT4, even with 13T training tokens, is 'undertrained'

#

i think they use tiktoken to tokenize everything

rancid otter
plush burrow
#

@vestal briar

#

Oh snap that actully my real name

#

in real life im not joking lol

novel plinth
#

Can anyone select a language on the ChatGPT website?

obsidian bluff
#

Anyone else having issues with GPT? I keep running into an error message.

rancid otter
#

btw i noticed chatgpt plugins disappeared on my plus account

wide pike
#

I don't use chatgpt that much anymore

#

so sorry

#

so Idk how i can help you

rancid otter
#

did they remove plugins?

#

oh wait i can just use the GPT anyway

ember nimbus
#

blud CANNOT code anymore

rancid otter
#

gpt died again

ember nimbus
#

did they try to change it or smthn to save some money

rancid otter
#

well it died for me

ember nimbus
#

ape brains running that if so like the only people that actually use chatgpt are developers or students to do homework they shd let it code at least

rancid otter
#

maybe its bugged in ur chat

ember nimbus
#

nah its all over the reddit

rancid otter
#

well tbh you didnt even use a prompt from what i can read there... just passed it stuff

#

its back here

ember nimbus
#

wym

#

i asked it to reroute the error messages into the error text box instead of console

#

old gpt could do it easily

rancid otter
#

wait u right. its at the end

#

yeah it does that sometimes

#

thats why you need better prompting

#

some ppl got good results by offering dog treats

#

or like 200 tip

#

its NOT a joke...

ember nimbus
rancid otter
#

i got muted...

#

yeah just use your prompt first like this:

prompt

`
code
`

#

also dont exagerate on the dog treats

#

wait no

#

its actually dying when creating the code

ember nimbus
#

im not roleplaying i just hand coded it

rancid otter
#

yes its bugged i tried your same prompt with GPT4 and it gets stuck

#

both GPT 3.5 and GPT 4 miserably failed

rancid otter
#

i wonder whats going on at OpenAI

sly geode
#

what model do they use when they post on instagram open ai ? for dalle 3 samples when i generate they arent even close to realism

sly geode
#

for images i meant

chrome bloom
#

Has anyone ran into the issue of DALL-E not creating images or logos based on their content policy? Simple prompts to create a logo keeps bringing up error messages?

dapper delta
#

Bye bye OpenAI, guys you cannot resolve issues for 3 days, this is recidivous, Google will crush you and you deserve it, you failed miserably

next stream
#

So disappointed in openai's handling of this issue

dapper delta
#

Yes, also, so unprofessional, they were super effective dealing with that spanish opera they had instead of dealing with the issues of paying customers

tight island
#

I am satisfied with OpenAI’s service

#

Five stars from me and will continue to use it

burnt hare
tight island
dapper delta
#

Paid bots which constantly spams the channel

glossy fjord
#

good day, I have problems, GPT stops listening to me and does what it wants

I write so that he writes the text next to him, he writes below him, I write it again and he copies the text according to which he should have done it differently
disobedience orders a GPT 3,5

dapper delta
#

Ok guys we know they gave you something to keep reminding us how great they are

#

Bots will remind you that its all good, you just need to pay it for nothing

glossy fjord
#

Anyone here using GPT 3.5?
or someone who has problems from GPT 3.5 refuses to obey his command x times?

dapper delta
#

I am using 4, it does nothing for 3 days now, any answer it returns is broken

glossy fjord
#

she started me with a again and again 3.5 GPT
and he repeats himself emotionally and refuses to listen and even when I point out the error he repeats it, I have the whole chat alerted to his error and he kept repeating the error anyway and when I answer that he shouldn't do it, it's done identically again and I don't know what's going on today, the bad GPT 3.5 may be terribly overloaded server?

dapper delta
#

Yea, also, same problems, I cannot assume what is going on technically on their side, overloaded or not, what is obvious was that they were super fast to deal with that internal spanish opera drama they had however not the same when it comes to dealing with problems of paying customers

glossy fjord
#

sad thing

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.

dusky cliff
#

have anyopne heard Nick Bostrom's today's tweet?

hasty vector
#

Does anyone have a problem using ChatGPT today? When I was using it today, the text stuck halfway through the generation. Clearing browser cache and settings and logging back in does not solve the problem.

next stream
hasty vector
#

Yes, that's the problem. Today I can't enjoy programming with GPT, right?

#

😭

zinc condor
#

I wish they have a message-refund policy. For every wasted message our count should ++

next stream
tight island
#

I have no problems with coding

#

Then again I use an actual customized gpt to my specific task

next stream
#

This problem can easily occur with java, vue or xml code

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.
tight island
#

But as I said unlike you guys I don’t use the main gpt 4

#

I use a CustomGPT specifically tailored to one specific coding task/Project

dapper delta
#

😀
Memories updated
The error message indicates that you are trying to use a Future The error message indicates that you are trying to use a Future
The error message indicates that you are trying to use a Future The error message indicates that you are trying to use a Future
The error message indicates that you are trying to use a Future The error message indicates that you are trying to use a Future
The error message indicates that you are trying to use a Future The error message indicates that you are trying to use a Future
The error message indicates that you are trying to use a Future The error message indicates that you are trying to use a Future
The error message indicates that you are trying to use a `Future

hasty vector
tight island
#

I just prefer the custom approach

hasty vector
tight island
#

It does just don’t start a new chat and copy paste stuff

#

Give it more information, how does Minecraft coding expressions work basic stuff, if you just start copy paste stuff with no information you get bad results

dapper delta
tight island
#

I recommend the Custom approach

#

If you have time to complain you have time to set up a CustomGPT for your project

#

🤷

#

I don’t deny that others are having errors, I just don’t get that particular error due to my approach

neat storm
#

^

#

Remember, videos are just a sequence of images played back at high speeds to replicate the illusion of movement

next stream
#

But it's definitely an OpenAI problem.

tight island
#

No doubt

neat storm
#

I wonder though

tight island
#

But at the flip side if you keep trying the method when it doesn’t work that’s a problem too especially when there are more methods

next stream
#

yeah

eternal rockBOT
neat storm
#

will Sora be able to generate images?
I mean, all it needs to do is generate one frame of a "video" and it's done
but that doesn't really make much sense

#

it'd be better to generate a full video and take a frame from that video

tight island
#

Ai Vid gens already exist

#

Out on the market ready to use

neat storm
#

can they gen images?

jovial root
#

I got the same problem, the chatgpt4 just have the text stuck halfway, it happen all the time. I have no problem with chatgpt3

tight island
#

Yup