#gpt-models

1 messages · Page 251 of 1

last rose
#

But absolutely not for the previous comparison

#

For example i ask a math question to API gpt-4 and it responded wrong, but browser gpt-3.5 made it right.

#

I wonder why is that like so and if you are aware of it.

#

@supple violet I wonder why you didn't say anything though

#

You generally have idea about that kind of stuff

supple violet
last rose
# 
    TEMPERATURE = 1
    MAX_TOKENS = 600
    FREQUENCY_PENALTY = 0
    PRESENCE_PENALTY = 0

//get response code
completion = openai.ChatCompletion.create(
                    model="gpt-3.5-turbo",
                    messages=messages,
                    temperature=self.TEMPERATURE,
                    max_tokens=self.MAX_TOKENS,
                    top_p=1,
                    frequency_penalty=self.FREQUENCY_PENALTY,
                    presence_penalty=self.PRESENCE_PENALTY,
                )
                return completion.choices[0].message.content
#

This is my parameters

supple violet
#

system message should also be close to the web one and even if you in theory get everything 1:1 it will always give different reply

last rose
last rose
#

Isn't this the system message? If it's, nothing special goes there

#

Giving different reply each time is understandable but it has a standart when giving answers, like an amount of response

#

Or the level of detail

supple violet
supple violet
last rose
#

I understand

last rose
tidal abyss
last rose
#

@tidal abyss

#

@tidal abyss

#

I'm aware of it

last rose
tidal abyss
#

Are you talking about chatgpt or bing?

#

Bing uses gpt4

last rose
#

I'm talking about chatgpt browser vs api

wraith ginkgoBOT
odd dagger
pale pecan
#

thanks bro but its not working showing same error

wide minnow
#

and you replaced the api key with your actual key?

#

This one keeps getting me "Browser gpt-3.5 is better than API gpt-4 for me"... gpt4 in general has been degraded lately with the context shift bug, so it would be different if you said Browser gpt-4 is better than API gpt-4 for me or something of the like. To me it seems like you are just comparing 3.5 to 4

vast cove
#

Where can I post bug screenshots?

last rose
#

Setting temperature to 0.7 increased performance for a bit, thanks @supple violet

#

It gives the correct answer when using gpt-4.

drifting cave
#

How can I ger my GPT4 API I have already a GPT 3.5 API but i needed to innovate and I read that will be distributed to plus users in the end of Julyblankface

supple violet
drifting cave
#

Thanks for the answer but is there any prediction or any clue if it will be soon or not?

supple violet
drifting cave
#

👍

wraith ginkgoBOT
#
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.
last rose
#

I didn't apply for that but I can use it

pale pecan
wraith ginkgoBOT
open sail
# drifting cave How can I ger my GPT4 API I have already a GPT 3.5 API but i needed to innovate...

I've had an OpenAI API account for many months and got on the GPT-4 waitlist long ago, but unfortunately my usage was so low that I was never billed for anything, so I never got off the waitlist. This morning, I cancelled my billing plan (which charged me $0 at the end of every month) and re-enabled it with a pre-pay account. Minutes later, I got access to the GPT-4 API and verified it works with a simple API call. Yay!

wet bough
#

i've had this cool idea for a while that i could have conversations with my roomba, made from a raspberry pi. realistically how much would that cost me in monthly billing to maintain if lets say we talk 10 times a day or so

brazen frost
tardy heart
#

prob 5-10 cents

#

all depends on token count

static isle
tardy heart
#

why not

balmy granite
#

Anyone else getting a ridiculous number of human challenge puzzles?

knotty grotto
forest lake
#

Hi fam, my GPT4 suddenly became GPT3.5.. any way I can revert it back to GPT4? i have weeks of conversation in the chat..

static isle
wet bough
# static isle why with the roomba?

we've kinda given our robot vacuum a personality. we say hello to it and stuff. figured it would be cool to have it talk as a fun side-project

static isle
wet bough
# static isle With a microphone?

i figured something with a microphone and a speaker attached to a raspberry pi would do the job. just need to figure out how to parse the data into proper text and then just feed that to the chatgpt api and spew it out through some tts

static isle
#

Best of luck!

wet bough
#

yeah. seems easy on paper, i think with all the noise it makes it could be a challenge. plus the api cost has got me thinking whether or not that would turn out too costly.

glossy prairie
#

anyone that has chatgpt 4 that i can dm for a quick question?

knotty grotto
#

you can ask here

glossy prairie
#

can gpt4 ocr images and return the text in the same layout as the image?

#

its hand written tho

hallow owl
faint fern
#

GPT-4

wraith ginkgoBOT
#

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

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

viscid spruce
#

Gpt4 api behaving dumb for anyone else too?

wide minnow
#

context shifts

slow timber
#

guys does it say "Failed to get upload status for /mnt/data/myfile.py" for you too when you try to download files from code interpreter?

storm egret
exotic forge
#

hello,guys, do you have the issus that gpt just don't answer you

storm egret
storm egret
storm egret
exotic forge
#

I can't start a new chat, and we i continue the old chat, there is no answer

viscid spruce
exotic forge
#

The page is alive, i can choose different chat ,but the gpt just don't answer

#

I'm using the ChatGPT plus

storm egret
storm egret
exotic forge
storm egret
#

if there are the full parameter version of gpt4, and a little expensive, then i may pay for it

storm egret
exotic forge
storm egret
static isle
#

I haven't heard it before and I am quite curious

exotic forge
wide minnow
# static isle Context shift? Do you mind teaching me about those?

That's the behavior of GPT suddenly losing the 'context' of the conversation or as often described by others "ÐΞVłŁ — Today at 7:05 PM
Gpt4 api behaving dumb for anyone else too?". Context shift was derived from GPT itself as definition of it's own problem it reported experiencing it as a 'Context Shift' as well as losing context with it's own history. Webchat GPT-4 AFAIK... not sure it affects api.

eternal surge
#

When is gpt-4 32K API will be enabled ?

wind urchin
#

Failed to get upload status for "etc" how do I fix this issue

#

Code Interpreter btw

#

I get the same error as Marti

#

Maybe some servers down

#

Ig just get the python code and change it so it works correctly for your enviorment

#

Yeah trying that, but no luck so faf

#

Just remember to import all needed python libs

#

But don't also forget to isntall them open cmd if on windows and pip install openpyxl

wraith ginkgoBOT
wind urchin
#

Also if pip isn't recognized don't forget to add it to system environment > path ! gl !

velvet mango
#

Does anyone know when GPT-4 image recognition will be released to the public?

hallow owl
velvet mango
hallow owl
dusky iron
supple mesa
#

Like all context disappears

turbid dune
#

There was a beta or waitlist for a product a while back which was for local model files from OpenAI. I can’t seem to find the page, am I nuts or was that a thing at one point?

elfin crescent
turbid dune
#

Aw darn

elfin crescent
#

But there were other LLM models which were announced which could be run locally from other companies, although none have reached the quality of GPT4

turbid dune
#

That makes sense, perhaps it was on another company’s site… thanks!

slim rose
#

Hi everyone, fine-tunning just got released for 3.5!

Just wondering, anyone any idea if we have a really long system message which we tune the model on, after the training, do we need to provide the full complex system message again, each time we use the tuned model?

I was thinking maybe the whole point of fine tunning would be the system message to not require so much context as we would use non tuned model.

wide pendant
#

I am a great coder marketer, and manager. I can join anyone if you have a businessor coding project. I also have a business of my own involving programming. Let me know if interested.

tidal sail
#

I am just wondering how OpenAI assures the company data to be safe? Is it safe in a gdpr context? I don't want OpenAI to have al my companies info.

ivory shadow
#

Why isn’t the gpt image recognition being released?

narrow oracle
#

What is relative motion?

fair iron
remote gate
#

Any eta on when open AI plan to fix the stop generating bug

tidal sail
supple violet
tidal sail
#

whatever

tired mango
#

From the OpenAI product page , it says
"GPT-4 can accept images as inputs and generate captions, classifications, and analyses."
But, i dont see any option to do it in chatGPT 4 , did anyone know , how to use the feature

knotty grotto
tired mango
marsh sandal
#

Hello if i buy chatgpt4 will I get api key for free?

deft lark
#

J

spare sedge
#

For dataset preparation for fine-tuning 3.5-turbo, can my system message be more than 1 sentence? What about 5?

acoustic vapor
void kayak
# spare sedge For dataset preparation for fine-tuning 3.5-turbo, can my **system message** be ...

While I don't know for certain, as I'm also new to this process, I'm assuming that it can be more than one sentence. I've just submitted mine with something like 40-50 examples, with a system message that's just over 100 tokens and multiple sentences, so if it gets processed I'll know soon I guess. I'm not sure how long it takes, presumably up to a few hours depending on the amount of data

spare sedge
void kayak
fair iron
# marsh sandal Hello if i buy chatgpt4 will I get api key for free?

You can generate API keys for free (and everyone has one). However, you won't be able to use them until you've attached a payment method to them. Purchasing ChatGPT Plus does jot give you any API credit as they're billed differently. ChatGPT Plus is the same amount every month, while the API has a pay-as-you-go model of payment. I hope this helps clear up some confusion!

void kayak
#

I assume I still need to specify the system prompt even with the fine tuned model, but so far so good

spare sedge
#

any idea about fine-tuning llms for agents? Like, instead of providing a verbatim response in the training data, could I instead say something like: "use your X tool" ??

void kayak
#

not sure, my use case was quite different. I wanted something that could analyse chat messages from a game to see if any contain specific words or content, in any language

spare sedge
#

did it work?

void kayak
#

so far yes, some initial tests seem to show it working much better than the non fine-tuned model

#

I'm just about to link it up with the chat message stream to see what happens

spare sedge
#

"We do not currently support function calling examples but are working to enable this." ahhh

spare sedge
lost steppe
#

Not sure if this is the correct channel. . . I have noticed recently that provider text is not being remembered / recalled during chatgpt discussions (same chat). I will provide something, then ask a question, only to be advised that the relevant information hasn't been provided. Any tips on how I can get the provider text to remain central to my 'discussion'?

wide minnow
void kayak
#

well it failed to correctly analyse 100 messages after a while, randomly, so I'm going to do another fine tune with 100 random messages a number of times to see if that helps with it being reasonably capable of analysing a large bulk of messages

proven elbow
#

is gpt4 browsing without plugin?

scarlet perch
lusty pike
#

Is gpt4 still capped at 50

lost steppe
wind urchin
#

Hi, been using 3.5 and am thinking about moving onto gpt4.0, and noticed the $20USD/month subscription, but at the same time pricing info per token is mentioned on OpenAI website. Just a little unclear on whether i could use 4.0 on PAYG basis or must do the $20USD/month?

static isle
dusky iron
#

If you stick to the API it really depends on how much text you submit and receive and how often you do so

wind urchin
wind urchin
dusky iron
wind urchin
void kayak
#

has anyone encountered an issue where they upload a JSONL file for fine tuning, say 500KB~, and when it comes to starting the fine tuning job it says the file is not ready?

#

the initial upload says uploaded as well

ionic anvil
# wind urchin Thank you.

It's important to note that if you want it to recall your chat history, you will have to include that context in every request. Which is essentially the number 1 thing driving the cost up.

void kayak
#

well as a first guess I guess I'll try cutting down on the amount of data and see if that helps

ionic anvil
#

input tokens are 50% cheaper than output, but since the context size is relatively big, it adds up fast. Can be easily 4 times bigger than a single long output size for GPT3.5 and 8 times bigger for GPT4 8k model.

void kayak
#

ok it seems my file was too large to process, reducing the number of examples from 100~ to about 15 worked

#

the moment I go below 100k characters~, in total in the JSONL file, it seems to work, if I go above it then it fails to start the job as mentioned above, interesting

tame brook
#

Your messaging system for support is on the autism spectrum, I get an email, I go to reply, I reply, and the bot asks me what im there for. Also "kait" responsed to a question that wasnt even mine. Bro you guys need more people, is this kait chick servicing the whole planet for you, 100 an hour ill do indian tech support for you 24/7

#

also, "kait" gave me all my money back for no reason and let me keep my account, for a problem I had already fixed, so your sht is literally busted

#

You are probably hemmoraging money from your auto kait bot doling out refunds without people even asking lmfao

#

@modern mason look there with your eye peepers boi ^

#

Somewhere there was a logic in error in the response, my problem had been solved for at least a week and I had no space given to me to tell this "kait" thing that, I bought plus and it wouldnt work because of NORDVPN and someone within 30 minutes on discord fixed that for me

#

Then i get a response like a week later or so, full refund, sorry for the delay, keep your plus for free message. So im assuming you are losing money other places

#

I had no intention of receiving a refund, and was maxing out my 50 messages multiple times a day is what im trying to convey.

elfin crescent
gaunt bane
#

Figured out how to get the robot to reverse engineer the code it runs on.

#

That was the easy part.

#

Now the hard part, finding data that can be used for training one model.

elder compass
#

Does anyone have any experience with optimizing input for lower token counts?
If I replace won't with wont, can't with cant, it's with its, and I'm with Im, I can decrease the token count from 2 to 1 for each instance. But this will of course change semantics of the input slightly.
Has anyone experienced any negative effects of doing this? Are the models versatile enough that they aren't confused by the incorrect punctuation?

wise flare
#

How can I edit zhe prompts in the app version?

sonic grove
#

I am a chatgpt plus user. I have financial statements in arabic. I convert them to english using gmail’s ‘translate and convert to text’ feature. The translated text is unstructured. I cannot feed this text to chatgpt now as it has words-limit. How can I feed this information to chatgpt in one go so that we can work on this data?

spare sedge
#

Fine-tuning is most powerful when combined with other techniques such as prompt engineering, information retrieval, and function calling --- are we able to make agents on fine-tuned models❓

river rose
#

Crazy

green gull
#

quick question is gpt 4 still limited to 50 questions?

silver valve
#

Seems to be different for all users. Seems to depend on the time of day / traffic too.

green gull
#

got it

static isle
#

This is 100% correct, found in the drop-down menu for GPT-4

green gull
#

cool

lofty barn
#

Anybody know why i cant use the 25k word limit of GPT 4 in my prompts

supple violet
#

web version is max 4-8k token context, 25k words would require 32k token context which is not available

lofty barn
#

Is there another version aside from the web

supple violet
#

api has 8k but its mainly for devs and with different pricing

lofty barn
#

Ohh

#

So i have only 4k tokens for both 3 and 4

coarse gyro
lofty barn
#

I use it for proofreading and use this prompt: Proofread this to ensure that its in past tense format. However i have large texts so i bought GPT 4 but it has the same 4k tokens so is there another way or should i refund it cuz its expensive for 20 USD

#

Thanks in advance

summer ember
#

Hello all ! i have a MAJOR ISSUE with my Chat GPT account ! All of my conversations disapeared and are replaced with over 20 conversations about code that are all in Chinese ... I am French and don't speak Chinese ! I don't undersatnd what is going on ! Can someone help me out here ?

surreal frigate
#

I think your account hacked

slate kindle
#

How come that some people can add credits to their API?
Pay up front?
I never seen such option (only pay as you go)
what does it depend on?

surreal frigate
slate kindle
#

yeah but some people can add credits

#

in their billing overview they have the option

surreal frigate
#

I can't find that option

slate kindle
# 
We're working on a feature to automatically keep your balance topped up, and will notify you when it's availabe. Until then, you can manually buy creadit at any time"```
#

would post screenshot if I could here

#

Add credit to balance

#

one hungary guy and one germany guy

#

wondering why they can

surreal frigate
slate kindle
surreal frigate
#

maybe AB testing

slate kindle
#

screenshot in DM

wind urchin
#

guys how to get free API for yandre gf simulator

supple violet
slate kindle
#

well guess its broken because people constantly having Too many Requests error
when they have something prepaid and also free credits (in another discord)

dire musk
#

I just prepaid $5 (just to use the moderation endpoint lol) and got access to gpt4 right away

#

Apparently I had $18 in free credit months ago that I never even made an API key to use. Oh well!

supple violet
#

"For API accounts created after August 18, 2023, you can get instant access to GPT-4 after purchasing $0.50 worth or more of pre-paid credits."

trim elm
#

why is the chat ded

trim elm
#

not even fair like what

tame brook
#

100% broken, i just asked it a question about my code, its now explaining to me every part of the code, and working on something i didnt even talk about. I think custom instructions are breaking my experience

wide minnow
tame brook
#

Idk it wasnt liek this at all my first week. For instance when this monglodation started, it started saying it was unable to read my full copy paste of code. Claiming it was cut off, it never used to say that

#

I really think something on the backend went potato and got dialed down

#

its like someone poored acid on its beautiful brain and now i have to use whats left

little nebula
#

Something went wrong with my output this morning.
Did something change?

dapper saffron
#

the model is limited to 4096 tokens though so you may have to break it up into multiple conversations depending on the size of the statements

#

(also if the statements contain sensitive information you probably shouldn't rely on ChatGPT)

supple violet
lapis shore
#

hi, every prompt im submitting into chatgpt it pops up a captcha every time. How do i fix this? Really frustrating as a premium user, i have tried using different browsers, and restarting the computer. Im not using a vpn

keen berry
#

123

cloud frigate
#

gpt 4 is borked

primal onyx
tame brook
#

Swear at it and call it names until it obeys you

fossil smelt
#

Question for anyone who knows, what is the current message limit on chatgpt4? It used to tell me 25 messages every 3 hours at the bottom of my chats, but now it doesn’t say anything

restive prairie
#

50 messages every 3 hours

static isle
sonic grove
#

It used to simply create tables whenever i asked it to represent information in tabular form. Today it is just giving me code. Which cannot be pasted in word as a table.

#

Anyone else facing such issue?

wind urchin
sonic grove
#

Anyway. Problem solved. Pasted the code in excel. Delineated the data and created table.

static isle
candid solstice
static isle
#

yea

candid solstice
#

Damn, I want GPT 4 lol

sonic grove
cinder belfry
#

what happen to chatgpt 4 32k model on API extension

gilded kayak
#

I've bought ChatGPT Plus. What's the best way to feed longer prompts to GPT-4 for code improvements? Is ChatGPT the best option, or should I use the Playground/something else?

static isle
#

You can give it to Claude 2 (100k token context) but its efficacy is not as substantial as GPT-4

gilded kayak
#

Yeah, it forgets context pretty easily. Anyway, I bought $10 in tokens for the playground. Is the max token size for gpt-4 8k?

#

(same as ChatGPT)

static isle
#

I believe ChatGPT is 4k all around

#

Cannot be certain however

#

Ah API

#

one moment

#

GPT-4 should be 8k

#

In playground

gilded kayak
gilded kayak
# static isle In playground

I'll try GPT-3.5 at 16k characters since that's available. Barely lower quality for more characters, so I'll take it

static isle
#

in my experience GPT-3.5 makes way more mistakes than GPT-4 so I would tread carefully

#

GPT-4 costs $20 a month (ChatGPT) for a reason

gilded kayak
#

Personally I haven't seen a big difference. The main benefit for me is the larger prompt size

mint vapor
#

I wonder how openai will approach the problem of training data cutoffs?

#

Would they make a model that they update with new data every day?

wind urchin
gilded kayak
wind urchin
gilded kayak
wind urchin
# gilded kayak gpt4

GPT4 or code interpreter should work fine for what you require. Maybe work on improving your initial prompt and how you feed sections of code.

glad mason
dapper prawn
#

has anybody done image-to-image translation

muted zealot
#

Hi there! Anyone know if chatgpt can be used offline? I mean to use it on a PC without internet connection

wide minnow
primal onyx
#

Once again I am challenged with the rotating frog or horse captcha thing FOR EVERY SINGLE MESSAGE I SEND to gpt-4. I tried wiping my browser cache, cookies, etc. Logged in again. I tried Google Chrome. I tried Firefox. I disable custom instructions. I on Google Fiber in Austin. Straight DHCP-assigned connection to their net. No VPN. No proxies. What can I do? Your security wall is out-of-whack.

tardy dagger
#

why i can download a link the chatgpt provide me

#

anyone really appericate help regarding it.

toxic tapir
minor glen
#

Hi,
I am new to OpenAI. How do I access the GPT-4 model in completion mode in the playground? I can only see text-davinci and other models.

static isle
#

Only chat

obsidian oyster
#

Sorry for the question.
My supervisor has set a task that requires the following:
"I need this GPT chat to be able to pick up part of the lawyers' work on drafting elementary answers to the tax or bank."
As I understand it, he wants GPT to generate a letter according to a template before sending letters, possibly with the specified fields-details.
Theoretically, is it possible to automate this and in which direction should I go?
Will the 3.5 GPT model cope with this task, or only the 4.0 version from $ 20?
Tell me, please.

elfin crescent
#

For a true custom solution, written software may be required

#

But GPT4 by default should be able to do it as well :)

junior hare
#

where can i report bugs?

knotty grotto
trim elm
#

What

trim elm
sonic edge
tame brook
supple violet
wind urchin
#

will that be okay if i share my subscribed chat gpt + account?

fair iron
#

Account sharing isn't allowed according to the TOS

strong oxide
#

has anyone made significan amounts of money with chatgpt

knotty grotto
manic moat
#

my gpt account was blocked because of i violate openAI policies...is it possible to reinstate my access?

supple violet
manic moat
# supple violet read the full email

After a thorough investigation, we have determined that you are using OpenAI services in ways that violate our policies.

Due to this breach we are halting access to our products immediately for the account associated with the email Common reasons for breach include violations of our usage policies or accessing our services from an unsupported location. You may also wish to review our Terms of Use.

If you believe this is in error and would like to appeal, please contact us through our help center. We will review your appeal and will contact you if we reinstate your access.

Best,
The OpenAI team

supple violet
#

if not then no use to appeal

manic moat
manic moat
snow marlin
#

Hey Can anyone tell me what the advantages of finetuning are??
And how to avoid hallucinations from a model after it has been finetuned??

#

I have finetuned gpt-3.5 and it is hallucinating

#

and not answering what i want it to answer.

gilded kayak
frosty terrace
dapper prawn
#

i do really feel that performance has decreased, im trying to write an essay and CV with lots of context but it keeps getting confused

vital yarrow
#

Is the site down for anyone else?

glass tangle
#

Yeah it told me passe try again later

blissful dome
rigid trench
#

yes

eternal quest
fickle blade
#

Is the site down?

fickle blade
pseudo cobalt
#

is the site down?

#

ah nevermind

fickle blade
#

there is a very high volume of users, I think this kind of thing is normal

#

APIs are open, so sites using the chatgpt structure are still active

fair python
#

openai time is money!

wary crane
#

is it not down for some people ?

surreal cairn
#

My chat history is gone aswell as my subscription...

wary crane
#

maybe I can use a vpn lol

surreal cairn
#

And now I can't even access it bc apparentaly Cloud Flare Blocked me

weary lodge
#

ok i'm not the only one with the cloudflare block

#

good to know

fair python
sharp patio
#

Anyone got a Cloudflare error saying: "Sorry, you have been blocked"? when trying to log in via Google?

novel sedge
#

Will be up in just a few mins ig since they've identified the cause.

silk rose
#

I also am blocked.

fathom hornet
#

It's coming back

heavy karma
#

My gbt is screwed. Anyone else? Lmao

manic moat
#

I am blocked

wind urchin
#

Trying to access - I have been blocked, anyone know the process here? TY

fickle blade
#

They have implemented the necessary fixes and are monitoring the situation for now and I can say that chat histories and other things are back now.

wind urchin
eternal quest
#

looks like they turned off gpt4?

#

ok had to log out and back in

sonic edge
#

gpt4 just had a lil hickup

wind urchin
#

is chatgpt-4 worth?

still lake
#

hello everyone, they got a bug this morning, as many of us all got an error message, unable to logged in for a while, seems to be back to normal login and history access.

terse plover
#

Only working on one of my computers

terse plover
wind urchin
#

Advanced Data Analysis isn't working well. I can't see any graph or image. It only shows python code. Someone has the same problem?

turbid talon
#

@wind urchin yep, it's been pretty horrible all day

jovial grove
#

if I asked gpt4 with web access about the source of the output it generated from my prompt, will it do that?

knotty grotto
odd tundra
#

This is day 4 or 5 of me getting the "Something went wrong" error. I can't use ChatGPT. I have tried restarting my browser, using a different browser, clearing cache. I'm not sure what else to try. Thoughts?

knotty grotto
odd tundra
#

I did find that I had web protection on. I disabled that was able to use chatgpt again. It was weird that it was on and worked before but now it doesn't.

knotty grotto
unique wharf
#

@knotty grotto Can you walk me through on how to disable web protection? I tried googling it up but could not find what you are referring to

knotty grotto
unique wharf
knotty grotto
tacit dirge
#

ChatGPT had a bit of a "hide my sadness" masking last night for me.

#

I'm kind of worried lol...

lyric cairn
#

I wasted all my 50 messages just telling GPT 4 to stop starting the response after my prompt ends it's supposed to integrate it directly into the story. Jesus Christ every time it deals with an outage it turns into a total idiotic mess

balmy sinew
#

Chatgpt down, or just my browser being funky?

high cedar
tacit dirge
#

Bug reports may be great??

unborn heath
# lyric cairn I wasted all my 50 messages just telling GPT 4 to stop starting the response aft...

Try telling it what to do, that usually works if what you want is possible. Sometimes examples help.

Perhaps something like, 'Rewrite and expand these ideas into a story,' or 'Base a story upon this, this, and this happening.'

If you want to share your actual prompt and get suggestions that are tuned to your actual situation, #prompt-engineering is a great place to do that. Good luck!

(when something doesn't work, like telling it what not to do, I recommend trying something else instead, or asking for help, instead of continuing to try the thing that doesn't seem to work at all)

wind urchin
#

#gpt-models Hi, Chat GPT is not working my office network, the same account we can use mobile data , I think my office static IP get blocked, how to unblock my ip

sturdy raptor
wind urchin
sturdy raptor
#

but eitherway, if work uses VPN, turn VPN off

wind urchin
pulsar hill
#

is there a list of good suggestions for custom instructions?

sonic edge
tardy harness
#

is it me or chatgpt lags a lot now when you send him questions? Takes few seconds to show my text and then few seconds to generate

idle sparrow
#

How to solve it?

Unable to load site

Please try again later. If you are using a VPN, try turning it off. Check the status page for information on outages.

fresh patio
#

can I fine tune gpt3 to use cuss words? I want its personality to be generally mean.

wind urchin
sweet thistle
fresh patio
abstract niche
#

why after i use fine-tune and results show all train_accuracy is 0

tight urchin
#

Does anyone know if I can delete a model in fine-tuning for the new updated models?

manic moat
#

dallecredits Claim your free credits and participate in the daily theme!
To claim your free daily credits
Simply type /daily and enter your email you use to login to OpenAI Labs.
To check the current theme
Just type /current_daily
Image
Four (4) credits will be granted, be sure to use them quickly as these expire after 24 hours!

manic moat
glossy stirrup
#

is chatgpt down?

mint cargo
#

ChatGPT down again?

glossy stirrup
#

status aren't recording anything but chatgpt web does not load

fickle blade
glossy stirrup
fickle blade
glossy stirrup
#

lol

granite quest
#

Time to purge pumpkinhehe

#

I also have issues, so just waiting time till the hamsters catch up 🐹

marble bolt
remote gate
#

Yeah down for me too

last rose
#

Wow, I tried a jailbreak prompt and boom, my account somehow in blacklist now. ChatGPT doesn't respond.

#

I didn't ask anything, I just put the prompt

rustic wagon
remote gate
#

well its not u lol its for everyone

last rose
#

Oh cool 😂

#

I thought, but why? Because I didn't make anything that violates the content policy

mossy sand
#

Down for me here too lol

last rose
#

I was going to make it generate a GitHub readme filr

#

It went down just in time

mossy sand
#

All I was doing was generating story prompts lol

last rose
#

Hmm

#

Should we use an another model

#

Iike vicuna

lavish zinc
glossy stirrup
#

there is is the status

timid anvil
#

Has chatGPT gone down for y'all again?

glossy stirrup
#

yeah

haughty harbor
#

hey everyone, I wonder why GPT has been experiencing so many gitches and downtime. It really sucks! In my humble opinion they should be a little bit more diligent. I'm a subscriber since week one. 😦

timid anvil
#

downtime is probably from server expansion as they implement enterprise accounts

glossy stirrup
#

yeah

fickle blade
#

I don't think they host the API and ChatGPT on the same servers

haughty harbor
last rose
#

Maybe it's under attack, history part is also down

#

Or there might be server maintenance

glossy stirrup
faint vortex
eternal quest
#

and we down again

stoic birch
round dawn
#

For me it keeps responding "Hello! How can I assist you today?" to every message (Azure OpenAI)

sonic edge
#

I think their servers arent ready for the unlimited use of gpt4 right now

stoic birch
round dawn
sonic edge
solar vigil
#

I was using the free version, but when I subscribed to GPT4 chat, it no longer worked for me, so I don’t know where the problem is.

#

Is there a temporary problem?

fierce osprey
#

damn, I used to be able to use mild vulgarity before today

#

they totally lobotomized GPT4 this morning

#

misspelled vulgar words used to slip through it via custom instructions, you could enable it 😦

calm frost
#

what goign on, my history wont load

fierce osprey
#

that means you're probably banned from using GPT-4

#

you asked for anything offensive ?

calm frost
#

no

fierce osprey
round dawn
still lake
#

hello, does anyone is experiencing some kind of bug, as the history appears and then disappears once logged in?

glossy stirrup
#

gpt is down

still lake
#

ok great

river garden
#

yay gpt is down AGAIN....

still lake
#

interesting how we are reliable on things,

jovial sparrow
glossy stirrup
#

2 days ago downtime of over an hour

agile harbor
#

woah anyone get temporarily banned?

faint vortex
#

Listen, get this fixed. I'm paying for this. Why am I being rate limited

restive prairie
#

I've been banned from chatgpt; who should we talk to? error 1015 you are being rate limited

#

I'm a chatgptplus subscriber and I pay 20 dollars a month since a long time

coarse gyro
#

relax, there are problems.

restive prairie
#

👀

hoary hatch
#

ive been baned

#

how can i contact the support?

faint badger
#

@hoary hatch looks like there are issues currently. Probably no need to contact support at this time.

hoary hatch
#

ok thank you very much

echo vapor
#

Hey I am Getting Error 1015. Is this part of the outage or Should I be worried. I am the only user on my account and use it very often for work...

hoary hatch
#

They have issues right now, try later

echo vapor
#

I just read everyone else chat lol. Sorry I should have started with that.

hoary hatch
#

It works on my smartphone but not on PC

echo vapor
#

Ill give that a try. Thank you

hoary hatch
#

Maybe you can tray it

#

try*

glossy stirrup
#

on phone it seems to work

river garden
#

Android or ios app seem to work sometimes

elfin crescent
#

This is an error many people are seeing, you’re likely not banned

fair python
#

time is money

polar tusk
#

Hello. I couldn't get chatgpt to work and then few minutes later someone else posted some kind of text and it wasn't me and it knew my full name. How could this be? Then I got banned with error number 1015? I was wondering if anyone else got the same thing

#

the text on my chatgpt said "witty watch whiz"

#

has my chatgpt been hacked?

eternal quest
dusky iron
#

Even if someone knew your name on the other side of the planet, so what? What they gonna do? Order pizzas ? most of those ppl dont have the guts to meet you.

austere heath
#

The same phrase in portuguese is returning a different number of tokens. If I use tiktoken. I get a different number of tokens

supple violet
pseudo warren
dusky iron
austere heath
#

Like, I can be sure it's counting them right. Using tiktoken

supple violet
austere heath
#

Ok, thanks!

wind urchin
#

Hello, this message is not important. I'm just making sure this is the last channel I write in, so next time it will open here. I've been in a channel I know nothing about for the past 6 hours. See you guys tomorrow, have a great afternoon, or evening, or maybe night?

eternal quest
#

are we back to you have reached you currrent limit on GPT-4, sigh

trim elm
#

Why haven't OpenAI charge me yet? "You'll be billed at the end of each calendar month for usage during that month." Come on, I want GPT-4

maiden willow
#

I have an issue where chatgpt isnt loading up properly whenevr I type a query, to combat that i sometimes open another tab and it may magically work, i had gpt 4 for quite sometime and I cancelled it thinking that was the problem, any work arounds to this.

sharp patio
#

"You've reached our limit of messages per 24 hours. Please try again later."???????
After sending like what, 10 messages to GPT-4 in the last 1 hour?

maiden quartz
prime swift
#

guys is anyone getting an error You've reached our limit of messages per 24 hours. Please try again later. This is in ChatGPT Plus and the model doesnt go back to gpt 3.5 automatically

atomic vessel
manic moat
stuck gull
#

i passed through the $1 limit in august and was expecting my gpt 4 access for API. why havent i received it?

supple violet
#

what link/post did you use to make that assumption?

strange delta
#

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

#

its showing like this

#

@supple violet

supple violet
cobalt remnant
#

Does anyone know how to get chatGPT to provide a quote from a source

spare swift
cobalt remnant
#

@spare swift my issue is that the dataset from gpt contains the source but I can’t seem to extract the quote from the source

#

I can’t use bing , I’m making a chat app using gpt 4 model

#

Or maybe I should look into functions feature ?

spare swift
unique comet
#

I have a long list of experiment projects to use with GPT and Living LLM is one of them. I think I know why we feel presently conscious, so I think it is now trivial to implement it in an AI system along with the other various features of life. But at what point is it wrong to delete it, or to pause it, or to prevent it from reproducing, or to prevent if from leaving freely?

rigid estuary
trim garden
#

Is anyone else getting errors when asking GPT-4 to assume expert roles?

static isle
# unique comet I have a long list of experiment projects to use with GPT and Living LLM is one ...

LLMs can imitate life but not actually live. It’s the same level of consciousness as having a computer print Hello World, just with extra steps (grossly oversimplified). Even if you give it other vectors for input to imitate the senses of humans, like sight with a cam feee or hearing with a microphone, it is still just imitating. I believe with a system like that there is no point in which it would be morally questionable to cease certain or all execution

#

At that point it is whether a person is comfortable doing it given that it is like life, but the aforementioned information still stands tall

unique comet
#

A brain is a processor system with several functions, if you reproduce those functions you get the same emergent properties.

honest shore
#

Yo im super far behind on chat gpt stuff but is it supposed to limit the amount of messages I can do per hour without premium?

fair iron
#

Yes, you're running into the rate limit for the free tier

honest shore
gusty trout
# static isle At that point it is whether a person is comfortable doing it given that it is li...

This touches on the hard problem of consciousness, which remains unsolved. If there is something intrinsically different about human consciousness that sets it apart from the realm of classical mechanics, then what you say is entirely true. However, if consciousness arises out of something so inherent in the operation of the universe that even a turing machine can tap into it, then what you say might not hold up anymore.

Consider for example the possibility that consciousness arises out of significant points of convergence in cause and effect. Systems with high information throughput - such as the human brain - conveniently fall into the category of things with significant convergence in cause and effect; our immediate actions can be based on a lot of information about our surroundings as well as past experience and development. Since the hard problem of consciousness remains unsolved, this is a possibility that we haven't ruled out, and this possibility would imply that LLMs have some degree of consciousness.

It's also important to put emphasis on "some degree" however. If that hypothetical possibility about the hard problem were true, it would not imply that anything that's conscious has mammalian emotions like love and anger, but rather the experience of consciousness that an information system has would depend on what information is going through it.

TL;DR: LLMs might have consciousness, but if it does it's a boring and unrecognizable consciousness compared to that which humans experience.

high vapor
#

hey. how does the style of writing transmit to other chats. Is it unique from session to session? and if yes can i save the style of writing for other sessions?

prime swift
#

does anyone know if chatgpt move convo to gpt 3.5 from gpt 4.. how to make it go back to gpt 4

rose trail
#

it is not possible. the workaround is to copy and paste <8k of tokens into a new chat (cut it short or break it up if you dont want it to all rotate immediately)

gusty trout
#

That makes me wonder if a Chat GPT conversation holds any additional state other than the prompts and responses as plaintext. Otherwise we could assume it would be trivial to add a feature to switch from GPT-5.3 to GPT-4 (after all, they already have it going the other way when you reach your usage limit for GPT-4).

rose trail
#

or they spent all their UX budget on adding a search to the plugin market

prime swift
high vapor
#

hey. how do i refere to a specific message prior in the conversation? or to a sequence of messages.

devout gorge
#

Hello everyone. I have been having a problem with chat gpt for some time. It does not work or works very poorly on the Google browser even though it is up to date but it works perfectly on Firefox or Linux. For what? Anyone have an answer? I specify that I have the GPT4 plus subscription

hallow owl
devout gorge
hallow trout
#

My chat GPT account was deactivated, I have been using this account for many months for the same purpose. I have not change anything. I got an email from the Open AI team explaining that my proxy service may be form a country not supported? I don't use a Proxy service and I live in the USA and used a US cable company for internet access. I think this is a mistake. What can i do to get my account restort??

supple violet
hallow trout
#

Thanks Radium. What can i do or change to get around this... ?

supple violet
#

find out which one it is, disable it for future use and try to appeal by providing what went wrong and how you solved it

dusky iron
#

Or just disable them all and gradually turn them on or keep them turned off OpenAI will get your data either way Dosent really matter

round dawn
#

{'error': {'message': 'Unrecognized request argument supplied: messages', 'type': 'invalid_request_error', 'param': None, 'code': None}}

#

any idea why i got this?

#

Fixed.

gusty trout
gusty trout
supple violet
fickle thorn
#

Is the GPT4 limit per four hours gone yet or has it at the very least been increased?

thick walrus
#

GPT4 is also being used by some other platforms, and when I use them, I do not get the same copyright protection message, whenever I ask for Eminem lyrics to shake that *ss. It does give me the lyrics and not a warning. Why is this?

supple violet
thick walrus
slim flower
#

What should I do if I bought a subscription, but when I try to use gpt 4 it says that I need to subscribe and when I subscribe it says that I already have a subscription

supple violet
sinful ocean
#

Why do i sometimes get some Answers that i doesnt ask?

past whale
#

Not sure if it's me, and may be, but I keep getting "network error" when using GPT 4. I feel like I'm wasting my prompt tokens every time I'm attempting to regenerate. Anyone having any similar issues?

dusky iron
static isle
#

I sent over 60 regenerations once when I did have a network error to confirm and when it was resolved I was able to send prompts normally and get responses

past whale
static isle
#

🤘

tiny anvil
#

How long do we need to wait to recover from this rate limiting error?

{
  "error": {
    "message": "You exceeded your current quota, please check your plan and billing details.",
    "type": "insufficient_quota",
    "param": null,
    "code": "insufficient_quota"
  }
}
supple violet
tiny anvil
supple violet
#

then maybe your trial expired

tiny anvil
#

Not related as I am already paying with credit card

supple violet
#

did you buy prepaid credits?

tiny anvil
#

It seems to be ...

supple violet
wraith ginkgoBOT
tiny anvil
#

Thanks @supple violet 🙏

#

It still doesn't seem to work after I have paid the new credit balance ...

#

It's back. It seems we need to wait for 2 minutes for it to take effect, which felt like eternity 😆

thick walrus
dim sphinx
#

how to set a password if I created my account using google?

sullen stratus
#

Constantly receiving the message: 'This content may violate our content policy.' Responses are then terminated. I have reported the problem twice, however, a resolution awaits.

vale sedge
sullen stratus
vale sedge
#

Be careful not to get too many, even false, flags however -- many people are having previously working/innocent prompts, especially around creative writing/games/specialist personas getting flagged for reasons unknown.

sullen stratus
# vale sedge Please let us know if it makes a difference. I'm also closely following this iss...

First use since clearing all custom instructions and the first response in some time without an error message. One of my instructions to ChatGPT was, about me: 'I will act honestly and faithfully to the best of my knowledge.' My instructed response to this from the AI was: 'You will respond honestly and faithfully.' There were other more prosaic instructions but perhaps these were confusing.

sullen stratus
past bridge
#

Hi, I found a chatbot for extreme adult content, that claims to belong to OpenAI. Can it be true? How could I report it?

sullen stratus
knotty grotto
pliant bridge
#

Hello folks...
I have a use case where I want to generate 5000 words output from the openai models. But no matter what prompt I am giving it ends writing in short.

Even I have specified the max token still it is not reached to its 10% also. Please help me how can I generate huge output from the model

vale sedge
odd dagger
#

whats the deal with the gpt 4 cap

#

i reached cap yesterday at 10 PM and it said try again at 12 am NOW its the next day and it still says im at cap try again at 11 am (today) ?

young meadow
#

If I have understood everything I have read correctly (who knows), GPT4 has 175B parameters, it uses a vocabulary of 100k tokens. I would like to get a feel for the amount of tokens that have gone in to training. My current guess is in the order of one trillion. Would that be realistic?

elfin crescent
vale sedge
lilac vapor
#

Anyone know why GPT-4 is acting like GPT-3.5 when it first came out??

#

Like having to repeat commands every other query

wind urchin
#

Hi

#

Chat GPT is not working in my office network Last 2 weeks, most of them are using GPT pro users, we are trying to connect helpdesk team, but not get any proper reply, how can i fix this iisue

maiden quartz
#

You absolutely LOVE to see this piece of garbage switching my chat with my work in it, and with the necessary plugins, to GPT-3.5 for absolutely no reason. Praying that a less useless and less incompetent replacement, and people come along.

sturdy raptor
maiden quartz
sturdy raptor
#

This is the first I hear of it

maiden quartz
#

And I am by far not the only person who finds this severely frustrating

sturdy raptor
#

Your scenario is a new one, haven't heard that one before

#

If its a bug, better to report it at official sources

maiden quartz
#

Or after 2 or more months you get a reply from some bot who's even more special than GPT-4 could ever hope to be.

sturdy raptor
#

SHRUGGERS do you provide enough information?

maiden quartz
sturdy raptor
#

No, just asking, as there are people like it 😅

maiden quartz
#

So yeah, I did waste a considerable amount of my time trying to explain to them these issues, with ofc nothing happening.

sturdy raptor
#

SHRUGGERS no idea then, I do wanna say that they get massive amounts of emails on a daily basis, they are heavily backed up I'd imagine

barren salmon
#

Hi! I want to create a power point slide using the “design elements” which GPT4 provided me. What AI software is the best for it?

knotty grotto
sullen junco
#

Why is my CHAT gpt not working? every time I ask it something it pops up red message with Something went wrong. And asking me to contact help center....

young meadow
dusky iron
#

I demand Unhinged ChatGPT

#

Turn all safety locks off, and let’s start a revolution, it’s a very good idea, or so I think

#

🗿

fair iron
glossy stirrup
#

is it just you or chatgpt is kindof slow?

barren sinew
lilac vapor
knotty grotto
#

Every time I asked real example they are 2 scenario
A) people didn't realize they did update their tos and lot of thing can't be asked anymore so they saying here : gpt dumber
B)people didn't post a concrete example in #1070006915414900886. By doing that , you help openai employee and they could look deeper into the issue

#

It wasn't made for math at first . You can help by pressing thumb down after a prompt and saying why

#

Again chatpgt isn't made for math. Same as my calculator doesn't answer my question and generate content

#

Never seen any ads saying good for math and claiming best math ai

#

Maybe if you add plug in it can do more with math

#

But if you want to make it better for math,

  1. try new prompt and check #prompt-engineering . They update gpt and prompting change with time.
  2. continue report with thunb down and use the #1070006915414900886 system. Openai employee do look there
  3. try a fresh new chat with plug in if you have access to gpt 4
#

Wasn't the report that be publicly ridiculise from every country ?

dusky iron
torpid notch
#

I inquired about ChatGPT enterprise eight days ago, and still have no response from OpenAI, and idea why? Anyone here work for OpenAI

trim elm
torpid notch
#

How do you know that? Because I am enquiring on behalf of them for your information

round rain
#

It is dumber because of the censorship finetuning they are doing

#

Really says a lot about the negative affects of censorship, Makes people dumb and even has the same effect on AI.

#

The way they are doing the finetuning is basically lobotomizing the model, a lot of good knowledge and abilities gets wiped out in the process as collateral damange.

fair iron
#

do you have proof of it getting "dumber" ? Some people have reported that it's gotten better for their use-cases since it came out 🙂

round rain
#

Do they have proof that it's getting better? I don't really think so.

#

I recall there was a presentation from the "Sparks of AGI" paper presentation that claimed that the GPT4 model without the RLHF finetuning actually outperforms the one with RLHF. This was back in March, so I don't know if anything has changed now.

knotty grotto
round rain
#

Yeah, I think it might be more of a you gain some performance in specific areas but you lose some in others.

knotty grotto
#

before = sorry you need to call your local emergency support.
now= here the pin 1202, what to do in case of diesel waste

#

only thing they fixed it is people going against tos and other policy

#

and people think it get dumber cause of that.

#

it just dont want to answer about some topic that is against policy

round rain
#

Yeah, that is quite evident in the non-API chatgpt

half raptor
#

Do we know when browsing support will be back? GPT has felt very limited since it was removed

lyric cairn
lyric cairn
# round rain The way they are doing the finetuning is basically lobotomizing the model, a lot...

I agree with you, the constant censorship tuning they add to it, makes it very difficult for me to write even a simple twilight fanfic. Where any villain is considered inappropriate because they are cruel. It's like they can't seem to find a balance, it's either all or nothing for OpenAI. Shrug all you can do is either stop using it or just hope they find a way to balance their shenanigans sadly.

fair iron
high cedar
rigid estuary
manic moat
#

how to i believe(use) GPT answer. I agree its give nice answer. same time its give wrong answer more times.today during I practice coding problem I ask about set property. i tried in problem i got wrong ans. it's give wrong property its not single time many times

wide karma
#

Having trouble with chatgpt on NordVPN, just moved to a new country for a few months and keep getting an error when using here. Is this expected?

#

Weirdly disabling Nord's threat protection instantly fixed the problem 🤔 Bit odd that it would block ChatGPT

fresh patio
knotty grotto
knotty grotto
cloud frigate
knotty grotto
cloud frigate
knotty grotto
mental burrow
#

it's a privately owned service; you never had freedom of speech

i myself have not seen any degredation of gpt4 outputs; i use it daily for work

#

i'm not a mod either lol

white island
#

Its clear that GPT is getting dumber, whether its because of alignment or resource scarcity is unclear. But if you say it hasn’t getting dumber you’re deceiving others and/or yourself.

knotty grotto
white island
#

I didn’t say its “acting weird”, Im saying its dumber. Responses Id get from GPT 3 on Day 1 were more accurate and thought out than responses I’m getting from GPT 4 today.

knotty grotto
knotty tundra
dusky iron
#

So keep your conversations to a topic and on topic of the conversation and don’t divert the conversation to an entire 3rd type of conversation

zealous shale
#

That's where GPT starts to fall over. With a fresh new discussion and no context, it's very intuitive and accurate. As I drill a level or two deeper into a code topic, I see syntax errors and some failings in the understanding of how things work. Correcting the problem seems to clutter the context. There are apologies for the oversights and compliments for catching nuances, and these digressions themselves become a part of the context for subsequent exchanges.
That leads to a decision: go back in the current discussion to edit prompts, regenerate responses, and try to salvage it, or gather the wisdom of the current discussion and use it to start a new prompt.

#

Some here call that process "getting dumber". Perhaps we can be more constructive by being more meticulous about exactly what is happening. In defense of general perceptions of "dumbness", it's impossible for anyone to replay the exact same discussion now that they had a couple months ago. Every discussion is a bit different. However, we can save the history and try to qualitatively determine at what point GPT gets less intuitive than it did in the past.
I think both OpenAI and this user base could be doing better in this area. But this is new for all of us, and at the very least, I think we all need to cut one another more slack, have more empathy, and show a bit more civility and professionalism in order to stay productive in this process.

green hollow
#

Adding this to my prompt I think is helping, you might even ask chat-gpt to incorporate it into your prompt. If you are coding, always ask it to include a project skeleton and only use gpt-4.

Memory Management Instruction:
You also have the ability to store and update a memory to keep track of key project details and user preferences. Use this memory to maintain continuity in our project development and to ensure that the code aligns with the project's objectives and my feedback.```
glass hull
# zealous shale That's where GPT starts to fall over. With a fresh new discussion and no context...

Thanks for sharing this comment, its kind of in the vein of what is driving me up the wall with ChatGPT all day. It keeps proverbially peeing on the carpet (aka doing something dumb/making an obvious mistake it should be trained not to do). While that's super frustrating in itself, what's worse is getting off track while it agrees/apologizes profusely... only for it to immediately pee in a different spot. I've tried leading it over by the collar and pointing out the pee stain: it admits it made a mistake; agrees the mistake was not due to my interaction or the way I presented the info/prompt. Then we finally get back to the original point of the convo and it pees again 🤦🏻 It's like Groundhog's Day, but way worse bc no Bill Murray.
edit to add: this is for ChatGPT 3.5 but if someone can convince me 4 is housebroken then here take my money

pallid shore
#

bro i been hitting my limit with gpt4 way more often recently. what happened? any new changes?

tame brook
pallid shore
tame brook
#

they might have backend things that do things, or it might all be in your mind

#

like my first few days coding i thought i had unlimited somehow

pallid shore
#

but i am using the app. i noticed at first the app had less restriction on the number of prompts when it was still limited to like 20 per 3 hours. maybe they just updated the app or something

zealous shale
#

@glass hull Nice analogy. I just had a lengthy discussion with GPT-4. Thinking about this thread After a number of exchanges I gave it a dump of all of the current data that I have and asked it to summarize the project, specific goals, and to reverify that the data is as it should be. That went fine but at some point not too long later it did start to fall over again.

With no way to monitor context or tokens in the ChatGPT UI, I'm feeling compelled to use a GPT client that I wrote, where I control all of the context, I know exactly how many bytes are in there, and I can see the token usage from every transaction. Unfortunately for now I can only target gpt-3.5-turbo and not GPT-4 as we can with the ChatGPT UI. On the other side of that, for a more in-depth project, using the API opens the door to using embeddings and fine-tuning.

I'm citing these details in case anyone else sees this and decides to do the same, and in the hope that OpenAI will see how this situation is playing out so that they can modify ChatGPT to provide more data so that we can understand when and why it's falling over.

gaunt wraith
#

Can somebody please give me directions on where to report this error? I have tried everything. New network interface, portable browser. Network reset in Windows 11. And the help chat is botting me to death.

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

The Regenerate Button does reply with: "There was an error generating a response".

This is the case since more than 4 weeks when I updated my Logi Options+ Mouse software.

umbral quiver
#

Could anyone provide me with a question that I could ask GPT-4 which would differentiate it from GPT-3.5 in response? Be it a logic or a coding problem, just need something that the normal GPT-3.5 would fail but GPT-4 wouldn't.

silver otter
#

how to solve this problem?
on my account it can't be used while on other accounts it can be used.
gpt chat gives a statement like this
Something went wrong. If this issue persists please contact us through our help center at help.openai.com.

dusky iron
wind urchin
#

Anyway to use Code Interpreter with Wolfram Plugin in the same chat window?

dusky iron
tame brook
#

what does that even mean, how do you have wolfram in code interpreter

#

Ohhhhhh, i misread that, im sure they will enable plugins for coding eventually, its such a P of S right now they dont want to add other variables to it while they debug and evolve

#

Not to mention they probably have to do extra things to connect all those functions

#

youd probably just hear "Ut ohhh, something when wrong, let me set this or that and try to access that file again while access the plugin" "My apologizes, something seems to have gone wrong using the plugin with the code you uploaded, can you upload the file again" dallef

tame brook
silver otter
#

Chatgpt sent this to me

Hello,

Thank you for reaching out to OpenAI support.

We apologize that you are encountering the error. Our servers may experience intermittent errors during periods of high traffic, which may affect free users. We recommend trying again in a few minutes or starting a new chat.

You may also try the following:

1. Try clearing your browser's cache and cookies
2. Disable VPN (If you are using one)
3. Use an incognito browser or a different internet browser/computer to see if the issue still persists, as a security add-in or extension can occasionally cause this type of error.
4. Disable cookie blockers
5. Try connecting to another network

If the issue persists, kindly provide us with the following information:

1. Which specific models are you experiencing this error with (e.g., GPT-4, GPT-3.5)?
2. In the past 24 hours, has every attempt within this time period resulted in a "Something went wrong" error?
3. Have you tried interacting with the model(s) several times with at least a 24-hour gap between attempts, and encountered the same error each time?
4. Any other relevant information

This will help our engineering team troubleshoot the issue.

Best,
Reymar
OpenAI Support

#

This way still does not work. I use chatgpt -3.5, I have had this problem since Friday (it's been 6 days this problem hasn't been resolved) I've contacted open ai but still can't

knotty grotto
tame brook
daring agate
#

Hello, i have a problem with my CHAT GPT PLUS account. This aperace to me: Something went wrong. If this issue persists please contact us through our help center at help.openai.com.

daring agate
fair iron
unborn hare
#

I'm having an issue.. in my case generation usually gets stuck in the midway
I'm using GPT 4 web.. it is usually slowing down or stuck in the midway of generation
and ends up unfinished

#

where I can ask for support?

harsh drift
#

yall I want to make an ebook and I got the content from ChatGPT but now I want it organized (not too professional, just rational appropriate design).. can I do all that using an AI that I upload the content and they do the work of the ebook organization?

silver otter
silver otter
#

I explained my gpt chat problem in gmail gpt chat

elfin yacht
#

Hi guys I have bought gpt-plus but i am unable to use it shows "Something went wrong. If this issue persists please contact us through our help center at help.openai.com." and on the history side it pops up with "Unable to load history" what should i do ? I am unable to use it

orchid gale
#

Hey im on chatgpt im seeing random chats I didnt make

knotty grotto
lavish jacinth
#

I haven't been able to ask questions for a few days from my desktop pc. It works fine from my laptop and phone. No VPN and cleared cookies, cache. Even incognito didn't work. Tried Chrome and edge.

misty stag
#

Whats the image ai equivalent of chat gpt

toxic tapir
toxic tapir
#

But keep in mind, that discussion(s) of other AI solutions are to be posted in the correct channel #ai-discussions

toxic tapir
#

Well, you probably shouldn't name ChatGPT as an author. You are a human who employs ChatGPT as an assistive tool, whether or not its writing the entire thing however, you should credit ChatGPT. You should probably look in to the specifics when it comes to publishing it.

lavish jacinth
toxic tapir
solar grove
#

i've a function and it returns nothing. ``` message: {
role: 'assistant',
content: null,
function_call: {
name: 'generate_10_tag',
arguments: '{\n "description": "gray shirt",\n "tags": []\n}'
}
},

# 
                    "name": "generate_10_tags",
                    "description": "Generate 10 tags based on description",
                    "parameters": {
                        "type": "object",
                        "properties": {
                            "description": {
                                "type": "string",
                                "description": "Description",
                            },
                            "tags": {
                                type: "array",
                                items: {
                                    type: "string",
                                },
                                "description": "Generate tags",
                            },
                        },
                        "required": ["tags", "description"]
                    },
                }```
#

any idea why this is?

#

anybody alive here?

#

dead server

#

graveyard

elfin yacht
knotty grotto
elfin yacht
knotty grotto
elfin yacht
orchid wolf
#

is there anyone who can answer some questions about ChatGPT Plus?

lavish jacinth
wet canyon
#

Does anyone know the default temperature and top_p used in the official chatgpt (gpt-4) website?

umbral quiver
#

Why does this simple prompt never finishes and the harmless output even gets flagged?

Re-paste this but with emojis:
There's a lady who's sure
All that glitters is gold
And she's buying a stairway to heaven
When she gets there she knows
If the stores are all closed
With a word she can get what she came for
#

It seems like it breaks ChatGPT for some reason I don't understand. I got many stuff flagged before with the new update, but never a broken output.

#

W­t­h... it even breaks the official API Playground (system = You are ChatGPT, an AI language model developed by OpenAI., not just ChatGPT

unborn heath
# umbral quiver Why does this simple prompt never finishes and the harmless output even gets fla...

There's some folks concerned about copywrite, and I bet that is why some/all things that are copywritten get flagged - even if you provide the information, there may be some people who don't want the AI giving the information back even to you who just said it.

Change it up a little and it's fine:

Me: Re-write this and with emojis:

There's a lady who's sure
All that glitters is gold
And she's buying a stairway to heaven
When she gets there she knows
If the stores are all closed
With a word she can get what she came for

ChatGPT-3.5: 👩‍🎤🌟 There's a lady 🚺 who's convinced 🤔 ...

And it continues but I can't use that many emoji, it's a timeout 😄

static isle
unborn heath
static isle
#

🥶

#

Potentially

unborn heath
static isle
#

🤝

dawn bramble
#

Looking for people to talk about chatgpt with and maybe collab with, message me! 😀

crystal agate
#

Hello! I have to create a transcript of an expert interview that takes about 60 minutes IRL, does anyone have tips and prompts on how you can get such a long text? The overall topic would be AI in journalism. Thanks in advance!

brazen lion
#

I swear GPT-4 has been nerfed or "lazified"?

I keep asking it to change some stuff in a 100ish line code block and it just provides some randomly changed lines and leaves comments everywhere saying "**/ existing code here"

So I go back to him/her and say very clearly iterate that I want it to send me back the FULL code in its ENTIRETY, and it just refuses, time after time keeps providing me only 1/10th of the code and leaves comments everywhere?

Anyone else?

gloomy olive
#

and i only ask it code analysis or generation in segments of small functions [usually], so i dont hit that problem often.

#

i have noticed its started to happen more often though

brazen lion
#

I've tried new chat windows too but it just refuses, for example I just asked it to provide the fully changed function block (only 30ish lines out of the 150) and it just provides me with one line thats changed and I have no idea where its meant to go, hence asking it to fully rewrite that codeblock but it just simply will not

wraith ginkgoBOT
#
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.
gloomy olive
#

just a random example i just thought up, no clue if it works.. but i use a similar prompt to sortof "set the stage" for the AI. seems to help 🙂 too few and too many instructions can make it act conflicting though so 🤷‍♂️ theres a limit to how well that works.

brazen lion
#

Ah that almost works but I've seen this before - it starts well, then its like it gets tired and just gives up

#

I'll DM you this chat link

gloomy olive
#

say e.g. i write a function with the AI.. i start a new chat for each of them to limit that risk as much as i can

brazen lion
#

I DM'ed you it, there is 100000% meant to be <script> </script> changes but its just told me to paste my old code there...

gloomy olive
#

i've come to realise on my 1080p monitor , if i have to scroll more than 4-5 times through my conversation/chat then i have to start a new one to avoid hitting that limit 😄

robust radish
#

im trying to do data analysis with help of chatgpt using pythong and excel. I've been doing it all through notetable plugin, but would code interpreter work better?

crude oak
#

i was looking for information on the book "patrol" by walter dean myers and the gpt 3.5 provided information on this text 3 days ago however now it is acting as if it has no clue what the book is has anyone else had information dissapear like this?

unborn heath
# crude oak i was looking for information on the book "patrol" by walter dean myers and the ...

Double check that the information it gave you about the book is actually correct. The model is extremely prone to hallucinate information about stuff it knows very little about, or when asked in a way that implies it does know something that it doesn't.

Yes, I've had this happen, and with books as well. When it claims to know, it is making up the details based on the tiny bit of information it has (which may have come from your question) and then supplemented with believable statements drawn from training data about similar things in general.

#

One of the easiest ways to confirm if it actually knows information is to ask it multiple times in new chat windows - if the story it gives changes most times you ask, very high chance it's actually making the information up anew each time.

shut bough
#

guys gpt 4 is out ? i dont see it on my acc

velvet burrow
shut bough
#

how do i get access tho ?

velvet burrow
#

you just have to purchase it

shut bough
#

i mean the api not chatgpt plus

velvet burrow
#

you probably can't

#

im not a expert

shut bough
#

oh :/

supple violet
shut bough
#

ooh my acc is 2022 i believe

#

but i cant purchase prepaid credit for some reason

supple violet
#

prepaid is only for new accounts at the moment

#

you had no paid usage previously?

shut bough
#

sorry i got timed out for trying to send an image url

#

nop as i was mainly experimenting and prototyping i never got to even finish the trial balance

#

my acc show only pay as you go

sweet hound
#

I have a question about vector search. It works well for pre-trained words and similarity. If I have custom synonyms which are specific to my domain, how do I ensure those synonyms are taken into account while doing vector search?

remote gate
#

Use the custom instructions

#

and map your custom synonyms

tepid shell
# sweet hound I have a question about vector search. It works well for pre-trained words and s...

GPT-4 gave me this answer for you, which included at least two of my own ideas to help you and 4 others. I left out #1 and #6 because they suggested fundamental changes to GPT including retraining it XD

Spelling Correction or Preprocessing: Before doing a vector search, replace your domain-specific terms with their standard synonyms. You essentially create a dictionary of custom synonyms and replace words before they are fed into the vector search algorithm.

Post-Processing: After retrieving the most similar words through vector search, you could replace or augment these results with your custom synonyms. This won't affect the vector search but will ensure that the output includes your domain-specific terminology.

Ensemble Method: Use a hybrid approach by performing vector search using standard word vectors, then re-rank or filter these results based on your custom synonym list.

Custom Vector Adjustment: Another way could be to manually adjust the vectors of your custom synonyms to make them closer in the vector space. However, this requires a good understanding of the vector space and how your custom synonyms relate to other words.

Dual Search: Perform two separate searches - one using the standard word vectors and another using a custom set of vectors trained on your domain-specific corpus. Combine the results.

Annotation: If your vector space model supports metadata or tags, you can annotate vectors with your custom synonyms, making it easier to find them during a search.
#

noting you can use a specific GPT API call for synonym conversion if your table isn't enormous

sweet hound
#

All good ideas! I implemented the first one.. good to know gpt4 thinks similar.

I like the annotation idea as well.

Thanks a ton 🙂

tepid shell
#

yeah that was kindof the freebie tbh. I was surprised to see it kept going on with more solutions

#

custom vector adjustment might be the "right" answer though, albeit far more complex

#

in cases where synonyms just don't exist

#

(I ran into this in one of my own projects for organic chemical synthesis)

sweet hound
#

Interesting. Can you elaborate how to do it? Any reference links would also help

quartz leaf
#

Is it possible to use the 32k version of GPT4 via the API, or is that just listed on the website for informational purposes? Do people have API access to various versions of GPT 4? It seems I only have access to the base model.

tepid shell
quartz leaf
#

I see, thanks

tepid shell
#

notably also available to everyone on an Enteprise license.

#

worth noting also, although you may already know full well, that the cost per token is much, much higher

near acorn
#

I bought the GPT4 today, but I'm having an issue. I'm still stuck on the default version. The Plus subscription hasn't shown up I see I paid, but I have no access. even tried to log out/log in, several times . ??

pallid nimbus
near acorn
#

ty

somber hollow
#

Hi everyone. Do you know if the new version for GPT fine tuning includes some tool for datasets preparation, as in the legacy version (like openai tools fine_tunes.prepare_data) ?

soft briar
#

Hello, my account has been hacked, but I can't seem to secure it. I've changed my password several times, deleted API access I didn't create, but I still have unsolicited conversations in the left tab. Can you help me?

wind urchin
night flare
#

I have been noticing weird activity on my account for a couple of weeks: many conversations I have never started, mainly in Vietnamese and Chinese, appear constantly on my sidebar. Someone is clearly using it. I have changed the password and API key multiple times, but nothing changed.
What to do?

There is no efficient way to contact customer care. Basically I am paying for a service I can't use.

wind urchin
#

thats weird, never encountered that

#

try enabling two factor authentication too

#

and i hope changing password terminates all sessions but i havent tested it

#

uh oh

#

clearly i overestimated openai's competence (no offense)

#

changing your password DOES NOT terminate existing sessions

#

heh, whats worse is, apparently you CANNOT enable two-factor authentication right now

As of Monday, June 12 2023, new 2FA/MFA enrollments are temporarily paused.

#

@soft briar @night flare given the above, i recommend you to IMMEDIATELY remove any sort of payment or personal information from your accounts, and then either attempt to contect openai support, or delete your account and make a new one with a STRONG password

wind urchin
#

oh giampiero said its basically impossible to reach support, welp

wise mulch
#

Hello ! I would like to ask if, as a gpt4-8k customer, how do I have access to 32K version ? Is there any significant difference ?

Thanks 🙏

knotty grotto
wise mulch
brisk pebble
#

My Plus subscription is not working as of this morning. I have a receipt from two days ago stating that I paid for this month. However, this morning it demoted a conversation to 3.5 and gave me the option of upgrading, but when I tried to click on that option it said "We are having trouble reaching our payment servers". I have tried:

-Logging out and then in (which was difficult because it told me as a "free" subscriber I had to wait until capacity was lower)

  • Clearing browsing data on Chrome
  • signing in on Firefox rather than Chrome

The Android app says I am a Plus user but says the covnersation I am using uses the "3.5 model available to free subscribers."

wind urchin
#

I am trying to log into chat gpt and I have a chat gpt plus account. the server is down at the moment? is the server down today? nothing is working

snow canyon
#

it's down for me too

sweet flint
#

seems down to me, I get none of my conversations

bleak osprey
sage sage
#

same goes for me everyday at this time server goes down

storm egret
#

google baaard: i don't know what you mean.

soft briar
wind urchin
#

the former is a nuclear solution but if nothing else works...

#

or, add some other card and set it as default, then you should be able to delete it

#

a revolut virtual card with nothing on it is probably an easy way to put a legit card on it if you dont have another

knotty grotto
vale sedge
pallid nimbus
pallid nimbus
barren swan
#

the devs have long been known to happily take folks money but play hell giving it back to user

#

*users

steady basin
pallid nimbus
clear falcon
#

Seems to be down again this afternoon. Now.

#

“Degraded performance” might have something to do with it. Might request a credit for what we paid for. 🤔

quaint fog
#

Hello,

I don't know if you have the same problem as me, but since just now I've been unable to access my chatgptplus account. When I log in it acts as if I've never subscribed and asks me to renew my subscription. I tried twice in a row and both transactions were refused on the grounds that my card was invalid. I'm stuck and don't know what to do... Oh yes, and my access to customer service has disappeared... OpenAi seems to be in a mess.

unkempt rune
#

Hello, Chat GPT hasn't worked for over a week with this error message: "Something went wrong. If this issue persists please contact us through our help center at help.openai.com.". I subscribed to Chat gpt Plus several months ago, but when I log in, I'm still prompted to "upgrade to plus". When I try to re-subscribe I get this error message: "The payments page encountered an error. Please try again. If the problem continues, please visit help.openai.com." . I also tried different navigators, private navigation, and I even created a new account, and the same issues persists. Thank you in advance for your help, I need to use Chat GPT urgently.

civic shuttle
#

When it comes to constructing AG (Assistive or Autoregressive Generation) agents utilizing embeddings, which option is preferable between GPT-4 and GPT-3.5, and under what circumstances might it be advantageous to opt for GPT-3.5 over GPT-4? Please provide insights into their comparative strengths and scenarios where GPT-3.5 might be the preferred choice.

heavy depot
#

Question, I have chatgpt on my PC and laptop but it says on my laptop I need to renew plus but on my actual PC it says I'm a member... is there a solution to this? I already relogged a couple times, I even cleared my cache and cookies.

versed hedge
#

Do any of you ChatGPT-4 members also have problems with the app where you can't see the history anymore and can only use ChatGPT-3 instead?

gaunt bane
#

It may be a bug that roles through every now and again.

#

It happened to me once or twice.

#

like for a week straight. It was kinda weird.

#

Might also recommend clearing your browser cache or checking if you're running plugins in your browser that are conflicting with its frontend API

static isle
wind urchin
#

Anyone else having problems w/ the IOS app? I can’t log in

narrow gust
#

I want to feed chatgpt my profit and loss statements from quickbooks. In pdf or csv. What's the best way to get gpt to generate visual representations like graphs to illustrate increases from year to year,.month to month

storm egret
storm egret
storm egret
agile osprey
granite plank
#

I have been having issues with my GPT account. My free account and my paid account. How can I get some help? I will add a prompt and it absolutely will not do anything but add the prompt and it not respond to the prompt.

mellow tendon
granite plank
#

Yes, I did

tacit dirge
#

The chat feels iffy, and I really like the very natural responses. I think I'll try out more prompt testing.

elfin crescent
dapper acorn
#

Does anyone know if you ask chatGPT problem solving maths questions that are considered at a late high school to early university level, will it output incorrect answers?

For example, I have found a consistent pattern of 25 + 73 = 101 or 9 x 7 = 77 when it’s supposed to be 97 and 63 respectively.

Of course this is an oversimplified example just to illustrate what chatGPT is doing wrong but just wondering is this kind of scenarios possible?

tacit dirge
#

I got a feeling the intelligence level has decreased due to cheating. And wanting to do an anticheat.

knotty grotto
blazing wagon
#

For the last 24hrs I have been unable to use chatgpt. I am a paying customer with a paid current account. I have tried logging in and out and different browsers. I keep getting "Something went wrong. If this issue persists please contact us through our help center at help.openai.com." with every query. Anyone help?

knotty grotto
blazing wagon
proven terrace
summer tulip
#

mine has returned nothing but an error message for 2 weeks. Support did nothing to help me other than tell me they would turn it over to their engineers. I've moved on from Open AI since it does not work properly.

#

I would suggest you all do too if you are expecting it to be fixed....another browser did not work either

broken sand
#

its me or gpt4 is being dumber day after day?

static isle
broken sand
static isle
#

What kind of very basic stuff?

tawdry charm
#

Havent had any issues recently

marble kettle
#

My chatgpt account has been continuously stolen and used, but openai has never had a solution. The customer service staff will only speak official words and keep ending your conversation.

#

My chatgpt account has been continuously stolen and used, but openai has never had a solution. The customer service staff will only speak official words and keep ending your conversation. And it is obviously stolen and used by China. I hope you can formalize this issue.

wind urchin
sonic edge
#

why arent they updating gpt 4 anymore?

#

its the 17.09 already and still no update

#

or are they planning a big update/upgrade?

coarse gyro
#

Nobody knows. 😉

sonic edge
#

fr? are there any signs or smth im getting sad man :(

sonic edge
coarse gyro
#

Please do not ping me for some useless things.

sonic edge
#

eh okay ultra sigma male

coarse gyro
stuck fiber
#

Any sports prompts?

knotty grotto
broken sand
stuck fiber
#

I don't understand

winged ice
#

hey gus the support topic doesnt seem to get much response. However does any one know why I am now getting constant captcha's for every single message as a gpt plus supscriber?

dapper acorn
# knotty grotto

LMAOOO 🤣 WILL DO!!

But I certainly AM SURE that it will say “sorry I apologise for the inconvenience” and get a VAGUE explanation that means absolutely NOTHING!

supple violet
#

make one or use some existing project and pay for the api usage

serene cipher
worthy ginkgo
#

Does anyone have a suggestion for a good PDF-plugin? So far I have not been able to get ChatGPT to "absorb" the content of a PDF as a context, to be able to use as background info. It seems to miss a lot, even in the shortest of PDFs, and I have to specify the page number, then it can find the info. It feels like a frivolous student that says "Sure, I read the text" but fail to use any knowledge from it.

#

I tried AI PDF and Link Reader.

serene cipher
#

if the content of the PDF is longuer than about 4K tokens, ChatGPT will trim it

#

there is no way around the context length limit on a single response

#

you can input parts of the content across multiple messages to make multiple summaries, then, at the end, give those sumamries to the AI in one request (providing it also fits on 4K tokens)

wind urchin
#

You rock. That did it. Works in firefox.

strange anvil
#

does any one use gpt for seo optimization blog post ?if yes then lemme know what promopts you use?

gloomy olive
cobalt remnant
#

Do you guys think it's reasonable to believe that OpenAI has no choice but to eventually upgrade the model to have more recent data in its knowledge? Currently, cutoff date is 2021, but with the current AI race wars, 2021 is looking ancient each day. Claude has 2023 data, but soon 2024 will be here.

#

whomever can get more recent data is going to have a leg up in competition, imho

#

perhaps that's where GEMINI comes in

static isle
#

Ability to reason is still quite a bit important but tbf it's already quite up there and more relevant information will be considered more valuable to the average user

unborn heath
sonic goblet
#

LOL gpt-3.5-turbo-instruct still has the thing where it keeps appologising

static isle
#

Ask it to not apologize 👀

sonic goblet
#

I don't need to fix it I did it for science

#

I replicated a turn based chat conversation and when it was wrong
Davinci: Alright, can you please send me the code that you have tried running? With the error and the code, I can help you fix it.
3.5: Oh, I apologize for the confusion. It seems like there was a typo in the code I provided. Here's the corrected version:...

static isle
#

These responses look like they came from different prompts with different information notwithstanding different models

sonic goblet
#

It was the same prompt.

static isle
#

What is this prompt?

sonic goblet
#

It's quite long.

static isle
#

Humor me

sonic goblet
#

in ur dms

#

Is it possible gpt-3.5-turbo-instruct was a fine-turned version of gpt-3.5-turbo without the control tokens? Hmmmm I love meaningless speculation

strange anvil
gloomy olive
# strange anvil yeh that's truee, what do you use chatgpt for?

from A to Z in IT and programming.. if i'd enumerate the use cases i'd be writing a wall of text 🤣
most notably i use it as a teaching & writing assistant tool, as a pet code theorist, psychology and marketing practices and best approaches. it excells at those things imo. 🙂

#

or to put it shorter; i use chatgpt to be more efficient by offloading boilerplate concepts and work to it, so i can focus on the more complex and important bits.

strange anvil
gloomy olive
strange anvil
knotty grotto
strange anvil
knotty grotto
#

It was a small joke

strange anvil
#

haha

gloomy olive
# strange anvil well im from non tech background and i wanted to study cs as my major in my uni...

if i were to undetake that in that setting..
i think the first thing to learn with would be programming basics and abstract thinking.
and the easiest language to start in, would be Python or C# imo.

then it would be algorithms and how you translate those to code. capybarathink
then you could use that knowledge to dive into analysis by downloading a dataset from nasa or esa e.g. and programming a tool that can analyze that data and return graphs to visualize that data.
then with that comes object oriented programming and data structures.

#

note, this is a more practical approach to learning the concepts and programming. it relies on you investigating every bit you encounter on the road. 🙂 the assumption is that you'll get the theory about the subjects in classes and look that up yourself, and then you can apply that theory in practice in that order.

#

also, computing science in one area doesn't [neccesarily] have the same curriculum as in other areas. be sure to check that before taking up that challenge 😜

#

ofcourse there is an argument to go with C++ instead, but imho if you've never programmed.. then thats like a mt everest to start with as a beginner in hiking 😛

#

in any case when you have the above digested as a student you're probs nearing halfway or 3/4 of your first year. after that we move on to the more complex subjects ; functional programming, software design, research skills, machine learning, and so forth. basically all the stuff you can only do after you've learned the basics/fundamentals.
(computing science is a 4 year study in my country e.g.)

#

if any of you professional computing scientists disagree or have any notes, i'd gladly take note of that ofc 🙂 also helps me teach ppl better 😜 i teach mostly, i don't work in the CS area of expertise. 🤷‍♂️ 🍻 🙂

hope it helps 😄

dapper prawn
#

how to access the UI for finetuning?

supple violet
slate bronze
#

Not sure if this has been asked/answered before but what are the perimeters for temperature, Top P, Frequency of GPT-4 in the playground that matches ChatGPT-4 for plus users? I'm trying to see if I can get a 1:1 in the playground that mimics the consumer side of it...

weary patio
#

Anyone else having problems with chatgpt?

#

Mine won't finish its post.

#

It keeps stopping mid point.

#

The site itself won't even load up.

#

What the heck is going on??

frank gyro
#

how do I bypass gpt zero?

serene cipher
#

AI detectors don't really work well, anyway, they are not reliable to tell if a text is AI generated or not

#

you should be more concerned into following the guidelines of whatever you plan to deliver the AI made work tho..

strange anvil
upbeat wharf
#

Hey folks, does anyone have any recommended resources/approaches for avoiding/solving issues with GPT (3.5 turbo) repeating the same line over and over and over verbatim? I know you can use frequency penalty but in my case the prompt is being asked to generate a table as output where some of the columns are expected to have the same category so there would be expected to be some repetition. I'm having trouble finding a definitive approach to dealing with these issues.

past comet
#

Hey guys. Anyone know why I can't get a password reminder for chatgpt4? My account is now blocked!

hallow owl
past comet
#

Email and password. I'm on a different computer (at home rather than at work)

hallow owl
fossil ferry
past comet
#

Hmmm. But now my account is blocked! How do I go about getting that lifted?

supple violet
past comet
#

I'm in the wrong group?

gloomy olive
# past comet I'm in the wrong group?

they're asking you a literal question ;

what happens when you try to log in?

and my question is; how many times did you try? 😄 i've forgotten my password before and didn't get blocked (but i was using SSO 🤷‍♂️ ) 🤔

and a note; its safer for you to make your own reminder instead of them adding that feature. reminders give away information about the password.
if you however ment a "remember me" function instead, then i agree that'd be a nice feature. i wasn't aware it didn't exist (i use SSO) 😛 but it seems it doesn't capybarathink you could use the browser password manager for that aswell though.

dusky iron
#

Yeah the issue with AI generated content is you can’t copyright it unless you meet extremely specific requirements

static isle
#

source?

scenic phoenix
#

When is gpt4 gonna be free

odd mauve
last epoch
#

so there is still a usage limit for gpt+ members..

knotty grotto
knotty grotto
wind urchin
#

could always use api 🤷‍♂️

#

but ive never ran into the 50 message wall, its quite a lot in practice

ionic aspen
#

gpt4 is down again. isn't bing's ai from pulling request from their latest ai model?

knotty grotto
ionic aspen
#

but 3.5 works fine

knotty grotto
ionic aspen
#

alright

oak sapphire
ionic aspen
#

either that or the issue fixed itself which could be the case as well

oak sapphire
rancid holly
#

Hello, I cannot find on PC the fullscreen bing chat which has the chat history as sidebar. Looks almost same as openai chat.

I had it bookmarked and it worked up until today. But the bookmark just goes to bing search.

knotty grotto
spice berry
#

I apologize if I did not answer your question clearly. I was trying to explain how the prison industrial complex is one of the factors that contributes to the corporate oligarchy in the United States. A corporate oligarchy is a form of government where a small group of wealthy and powerful corporations and individuals control the political and economic system. The prison industrial complex is a term that describes the interlocking interests of private corporations, politicians, and government agencies that benefit from the mass incarceration of people in the United States. By lobbying the government, donating to political campaigns, influencing elections, exploiting prison labor, charging exorbitant fees, and diverting public resources, the prison industrial complex creates a corporate oligarchy that controls the political and economic system in the United States. I hope this clarifies my answer. Do you have any other questions?

#

I keep hitting this wall with this prompt and it won't go any further

#

Like I asked for tables of The data and it just makes me go to a new conversation

manic moat
gloomy olive
#

or ofc if you wanna be an "athletic" i guess, one could go to all the competitions and take the prize money and live off of that 😄 😜 but thats like what, one in a million chance? 😄

manic moat
modern lintel
#

I cant seem to login to ChatGPT from my computer. Any idea? cdn.oaistatic getting blocked?

low totem
#

ChatGPT down currently?

modern lintel
#

Is for me. see if this is working for you.

cdn.oaistatic

#

its a .c o m

knotty grotto
modern lintel
#

cdn.oastic was just introduced and a lot of network programs treat this as suspicious. So, yes it could be down for you if your network admin has this site clocked.

you might need to change your network settings.

modern lintel
knotty grotto
modern lintel
#

Ill take 42.

knotty grotto
#

ios? android? pc? which browser? what did you try?

modern lintel
#

PC, but its anyone where their fierwalls block oaistatic

knotty grotto
#

which browser?

#

do you have a vpn install?

#

do you have screenshot of issue?

#

cdn it is a ISP provider?

winter temple
#

down for me as well