community-help

[stiff ruin]

Does anyone have a recommendation for how to get a close up, centered shot of an object consistently? I am only averaging 1, maybe 2 correct images per generation. Any help appreciated.

My current prompt: "Close-up, center of frame, [object], [art/photo reference], DSLR, HD"
I have also tried "perfectly centered" with similar results.

[gaunt canopy]

hello, I recently found that a lot of times, when creating a new image based on a previous one +mask (i.e., "edit" mode), I get a lot of artifacts in the form of strange colors and images that are not "as beautiful" as the ones I can create from scratch (i.e. DALL-E images in 'create' mode, not based on a previous image). Any idea why it happens and how I can prevent this?

[verbal crag]

You might want to try this: #❌┃tips-n-tricks message

[verbal crag]

The "strange colors" you see are artifacts from the original image. When outpainting, make sure to erase the artifacts so the new image would not be influenced with these strange colors. Or you can ask for a muted color palette, the editor guide states the following:

Ask DALL·E for a muted color palette, especially as you stray further from the center, to avoid oversaturation and color-blasting.
https://help.openai.com/en/articles/6516417-dall-e-editor-guide

[gaunt canopy]

thanks a lot! I will try these

[placid roost]

Why does Dall-E make such a visible line between the original and generation in Outpainting? None of the 4 versions work out correctly

[zinc marsh]

If I recall correctly, DALLE has (or at least had) a slightly bad habit of raising the saturation on each successive outpaint frame it generates, which could be leading to artifacts like that; I've often see people add "muted colors" to their outpaint prompts in an attempt to combat that. (I suppose there could also be something in your prompt unintentionally incentivizing the AI to do that, but that's a bit hard to judge without knowing the prompt you used.)

[willow mural]

Blending/feathering tools in a photo editor could also help remove this seam

[placid roost]

The prompt was copied straight from the other AI generator I used

[placid roost]

Yeah I had to do manual polishing to make the picture look good

[willow mural]

That's typical in my experience

[regal lynx]

All image Generators are trained using different images and are processed differently. You can't expect the same prompt to create the same type of image from one generator to another

[willow mural]

Until dall e has a consciousness we can give feedback too it'll never be exactly what you intend

The ai edit feature is good

Though i wish there were different eraser sizes and shapes

[regal lynx]

You can change the eraser size! There's a little slider on the right side of the screen if you're on the desktop website

[willow mural]

I've only used the desktop site for one session so I must've forgot

Typing prompts in on my phone randomly on breaks at work is great fun and gets ideas going for my own work

Does anyone know if uploading a photo with transparency will let you edit it right away or do you have to erase the pixels in the web app

I know transparency uploads work on the outpainter

[willow mural]

Turkey erasure. Typical theme for today

[wraith warren]

You should be able to just upload the image with transparent bits and you won't have to erase anymore. You will just need to make sure the generation frame covers the transparent bits

[willow mural]

I meant in the single frame editor, I can test it sometime

I sure hope they do some kinda mobile access to the outpainter soon so I can do stuff on the goes

[sweet plank]

I've had this question for a while but I couldn't find it answered in this server but do the and the ⭐ both have the same function?

[rustic marlin]

yes

[regal lynx]

It was a recent generation I made 😅

[past comet]

Some person claimed they "Trained the ai with images to help the artstyle". Is this a feature, and if so , how do you do it.

When i asked them what that actualy is

they said they used "An upload image to edit" thing

but the problem is, that expands the images and refrences the images, not just refrencing them

I am generally looking for a method of sending a refrence image to generate an image with no remnents of that refrence image

[regal lynx]

Dall-E 2 doesn't offer this function. You cannot "train" the model in any way shape or form

[past comet]

Oh, thanks for notifying, i am very confused why they TOLD ME they did

but whatever

[rugged wing]

Someone from OpenAI told you that? Or just some rando?

[past comet]

A random person

they probably just missunderstood

Smth about the image editing

[dusty basin]

i have a general question, because im new, how do i make the ai use my face for generations, the takeyourprompt function??

[golden ruin]

You mean adopting? You can upload your picture and do modifications but there's no specific command to use your face while generating

[dusty basin]

ive seen a science yt who used his face to make ai generations

maybe its another ai but he said is dall e 2

[dull basin]

[❓] — Q: Can DALL·E transform the style of my image into another style?

We currently don't support transforming the style of an image into another style. However, you can edit parts of a generated image and recreate them in a style you define in the prompt.

[split hazel]

I don't think that's Dall-E 2

[golden ruin]

It needs to be specified which part of this illustration was made with Dalle2

[rugged wing]

Hey does dall e 2 use python or js(or something else)?

[golden ruin]

Do you mean if you can use dalle2 in for python or js?

[rugged wing]

Oh nvm

[dawn lake]

Greetings to all! Can someone tell me how to get the AI ​​to display text exactly on the image? In this case, I would like to somehow display the word ELPIS on an image, but AI cannot match the order of the letters.

[dull basin]

[❓] — Generating texts

DALL·E 2 is not engineered to properly generate text. While it may get a few simple words right now and then, that is the exception rather than the rule. If you need to add text to your image, consider using an external image-editing software.

[lucid ravine]

The simple answer is you cannot prompt for cohesive text in DE-2

You may get lucky with multiple attempts but there isn't a single way to do it that works consistently

[unreal gorge]

You can fix this in post-edit, but you can greatly reduce if not eliminate this issue if you use the eraser tool prior to the outpaint gen to ensure you don't have hard edges. I never leave the original border as I outpaint and rarely do I ever see a seam.

[rustic marlin]

I think it is, a composition which used dalle at some point

[rustic marlin]

we don't know what the OpenAI's backend use, but there are libs for both of those languages for the API

[rugged wing]

Ok

[dawn lake]

thank you. In the meantime I got lucky and AI hit the correct order of letters and made a great logo. 🙂

[lucid ravine]

woot!

[patent hollow]

Nothing quite about how to use the application, but does anyone know if you can use the image it generates for things such as images for cards?

[lucid ravine]

Yes.

[stone pagoda]

Does anyone know how to generate specific color theories into the generations? For example an analogous or split-complementary with the primary color red. Thanks if you know and share

[regal lynx]

I recommend reading over #1021130377026351105 tip: #4

[zenith jungle]

hey, I can't download my image for the outpainting, is there a reason why ?

nevermind, i figured it out 🙂

[keen hamlet]

Checking for specific color theory terminology is best done via direct testing. Be aware that there is an interaction between the subject matter of the prompt and any art style cues, including color, so a base prompt needs to be crafted that is unlikely to interact with the color testing. Prompt Tip #4 uses abstract objects for that reason: https://discord.com/channels/974519864045756446/1021929896185970708

I am not even sure how to do testing for color sets based on theory. Here is a fail: "Red, Yellow-green, Blue-green split complementary color wheel with the primary color red." The limiting factor will be training data, which is probably not tagged with color theory terminology.

[terse tide]

hi, im already a user of dall-e, how can i use dall-e 2?

[golden ruin]

https://labs.openai.com/

[dull basin]

<:dallelogo:1000465618258100294> You are using DALL·E 2

The invite you received from OpenAI is for DALL·E 2. The first version of DALL·E, version 1, was never released to the public.

Read more here: https://openai.com/dall-e-2/

[uncut chasm]

What Is OpenAI codex

[sweet plank]

It's an AI that created code based off of natural language

[regal lynx]

It's a bot that converts natural language to code

[uncut chasm]

Like how does it work

[regal lynx]

I think OpenAI pours fairy dust on GPUs and they do magical things. I have no idea how the backend works

[uncut chasm]

also why does everytime i use it it just gives me a image error

nvm

it works and its good

[vestal stratus]

is there a reason why gpt3 only knows things before 2020?

[regal lynx]

The dataset it was trained on was likely made before 2020

[chrome swan]

I'm trying to generate images of people training jiu jitsu and they're coming out quite nicely except in every image there is at least 1 extra hand, arm, leg or other random body part, other than what should be there for two standard humans 🤣 Any ideas on how I should instruct differently?

[regal lynx]

I would recommend inpainting the images to delete the extra limbs

[chrome swan]

Hmm ok, I'll try

[minor geyser]

why was entering the server such a hassle? The button was saying "This Interaction Failed" over and over again. I probably clicked like 200 times just to get in the server.

[regal lynx]

We are aware of the issues. With the overhaul, we've unintentionally introduced a couple bugs. I believe that the server is syncing with a previous version of it currently that will apply the verification role to all previous users. Sorry for the inconvenience

[split hazel]

The bot was struggling with too many requests for a little bit but should be fixed now

[keen hamlet]

Every word in a prompt affects the results. To suggest improvements, you may want to give the full prompt and a full set of results in #❌┃prompt-help.

[olive shell]

I was wondering what the special role mentioned in the announcement is?

[regal lynx]

It's the 🥑 role for users that were in the server before the shift

[lone laurel]

Hi
if not difficult tell me how to use Dalle-2, just for some reason it is not available in my country

[lucid ravine]

@lone laurel Hi, did you already try signing up through https://labs.openai.com/signup ?

[lone laurel]

Im from Ukraine if it matters

[lucid ravine]

Ahh. I'm sorry to hear that.

You'll need to contact support to get insight into the situation. [e-mail]

[lone laurel]

why is Dalle-2 blocked in some countries?

Or is it something else

[rugged wing]

Now that the DALL-E Discord server is the OpenAI server, because it encapsulates all products released by OpenAI, will there be channels for Whisper? I see some created for APIs, for DALL-E, for GPT, but not Whisper.
Or is this still in progress?
Thanks.

[trail edge]

Question, I have used my phone number on accounts for family members on DALL-E and now want to do it myself but it says my number has been used too many times. What should I do? I only have the one number haha.

[regal lynx]

A phone number should only be linked to one account. Linking the same number to multiple accounts breaks the Terms of Use

[unreal hemlock]

<:book_icon:1046291503158399107> `` Rule 0 `` Adhere to the following Terms & Policies:

Discord - https://discord.com/terms
OpenAI Terms of Use - https://openai.com/api/policies/terms/
DALL∙E Content Policy - https://labs.openai.com/policies/content-policy
Additional DALL∙E Terms of Use - https://labs.openai.com/policies/terms

[trail edge]

Interesting that it allowed it. Is there a way for me to close those accounts and move forward with just one?

[regal lynx]

Contact [e-mail] about closing accounts please

[trail edge]

I will, thanks for the speedy response!

[trail edge]

Do I need to provide any info in the email? I imagine the phone number in question.

[regal lynx]

I would recommend explaining the entire situation and the OpenAI accounts linked to the phone number

[sullen swan]

Does OpenAI plans to make an AI video generator

[rugged wing]

I Need help

This open ai playground doesn't work for me at all

Omg this open ai needs so much work

It says I've reached my usage limit

I haven't even used it

And I swear you get tokens each month?

I need help

[rugged wing]

Ok I found my issue

The 3 months thing is over

[winter elk]

what is gpt?

actually no, wrong question, how do i use it?

[strange mesa]

Could someone help me a tad? I am trying to generate seamless camo, is there some specific term that I should be using? Results I am getting are nowhere near what I am trying to make, here is the prompt I was using winter digital camouflage and the image result

[visual wren]

It's blocked in my country too :P ||I went through a lot of hoops||

I found this thread about it https://community.openai.com/t/why-is-openai-api-not-supported-in-certain-countries/15395

[scarlet idol]

What do you mean by "camouflage"? If you're trying to get a military clothing camouflage, there are quite a few common patterns ("leaf" camouflage being common in the US).

You might want to add "clothing pattern" to your prompt, too, after the name of the camouflage pattern.

[near pollen]

Hi!
Is removing image from history or hiding it there possible? I played with very abstract, broad prompt and it generated one image I need to get rid of.
It is not breaking any rules, probably wont be harmful for anyone else, but it is triggering my PTSD symptoms, flashbacks.

[split hazel]

Hi, I'm sorry to hear that. Removing / hiding images from your "All generations" history isn't possible at the moment. I'd suggest contacting [e-mail] about your situation and the possibility of a removal. Hope this helps and I hope you're offered a solution.

[near pollen]

contacting support sounds like possible solution. I thing will also use suggestion form, and... play with AI a bit more without "abstract bad prompts" like one generated my trigger. It is ironic that I typed: "The funniest thing in the world".

Also, ublock is working till page reload, I tried it before asking on discord.

[split hazel]

Apologies for the deleted message, kindly disregard its contents as it may not be as straightforward as it sounded. I noticed it's not working on page reload, which is strange as I'd expect it to.

Ok, I have uBlock working now, it won't revert on page reload anymore. You'll need to remove everything after image.webp in the element picker before proceeding.

[near pollen]

No problem. 🙂 I also expected it to work - because it works on most places on internet. Also... this image might be actually really useful for my therapist, I just emailed it.

And removing everything after webp (with part of screen hidden behind my hand) resolved issue - I do not have to see this thing ever again. Thank you so much. 🙂

And it is still ironic and, in morbid humour way, funny that "The funniest thing in the world" prompt generated it.

[split hazel]

Glad I could be of help

[gentle dock]

Sorry for bothering but could someone tell me how many accounts can be made with the same phone number?

[split hazel]

I'm unsure of the exact limit, but I believe only one of the accounts will be granted the $18 API trial and recurring DALL·E credits. Circumventing this breaks the terms of use.

(e) Free Tier. You may not create more than one account to benefit from credits provided in the free tier of the Services. If we believe you are not using the free tier in good faith, we may charge you standard fees or stop providing access to the Services.
https://openai.com/api/policies/terms

[gentle dock]

Perfect thanks for the advise

[toxic gale]

The bot Dallify shows it in 7 servers. Does anyone know how I can add it to my server? I am assuming I need to put in my own api keys and that is fine.

[regal lynx]

I recommend asking in #dev-chat 🙂

[rustic marlin]

the bot is open source, if you know TypeScript, you should have no problems to run it
https://github.com/openai/dallify-discord-bot

[potent lichen]

hey guys.. when it says the account was blocked after too many tries to login (i couldnt recall password, but i know it now, one try late).. is there any way to unblock it? or is it blocked for like an hour or something?

[rustic marlin]

did you figured it out yet?

i don't know the answer, im just curious if it unblocked you already

in case it didn't, try recovering the password via email

[potent lichen]

so far it didnt

i am waiting for the support to respond

@rustic marlin

[rustic marlin]

damn.. well, support should solve that indeed

[potent lichen]

hopefully, yeah

[glacial charm]

Hello I have a small problem, why after clicking on the validation arrow nothing happens the same with the trash can

[rustic marlin]

the script of the page probabily crashed

and not it is not responding to any input

try downloading the image and refreshing the page

[glacial charm]

I tried to download the image but it didn't work.

[tired hemlock]

Is there any open source implementation of OpenAI's Dalle editor ? I need to integrate this to my website but hard to find one.

[regal lynx]

I don't believe the editor is open source. @rustic marlin will know more than me though!

[indigo rune]

Not officially but maybe there’s community versions

[tired hemlock]

Yeah I thought so but any similar alternative would be fine.

[indigo rune]

There’s this sd one https://github.com/lkwq007/stablediffusion-infinity

[rustic marlin]

it is not

[tired hemlock]

Actually it's hard to play with canvas and stuff.

[tired hemlock]

Yeah I tried this one but it's gradio based demo and isn't similar to what I need. I wanna use an editor to integrate on web as I donno much about react etc so that I can send req to Dalle via an Api.

[thorn delta]

Is it just me or does DALL E struggle with mythological concepts? I have now tried Cerberus with every prompt I could think of and can only get terrifying dog amalgamations

[rustic marlin]

it struggles with things that are too unnatural, for example, creatures with many limbs, heads, animal body such a centaur...

[rugged wing]

It can also struggle with the number of limbs on natural bodies. 🤨

[fickle kelp]

Perhaps try editing a good picture, erase part of the shoulder and prompt it to add another head?

[odd gale]

Hi All! Is there a way to hook OpenAI with zapier and tweet - for example one paragraph - by a certain schedule?

[mental juniper]

Maybe a kron job

[rugged wing]

How do I see my api key? For me it is just showing sk.. then numbers but not the whole key

Lugui am I allowed to @ you?

[rugged wing]

Bruh I straight up can't see my api key

[regal lynx]

I believe you'll have to delete your current one and generate a new one

[rugged wing]

Aw man

Ok

[regal lynx]

After you first generate it, it disappears to being completely shown (this is my understanding)

[rugged wing]

Screw security. I made a new one and saved to to Google drive

I haven't linked my card so I don't really care

[high trench]

In a 27 November announcement, there was talk about new discord roles being added for developers and enthusiasts, how does one go about getting the 'developer' role? Thanks in advance.

[split hazel]

#server-roles

[high trench]

oh, haha, i didn't even notice that channel. 🙂

[empty tiger]

When will Dall-e support Paypal or iDeal? I want to buy extra credits, but I don't have a credit card...

[unreal hemlock]

<:book_icon:1046291503158399107> `` Rule 3 `` Stay on topic.

Be mindful of what other users in a channel might find helpful or interesting when posting. Stay on topic in order to keep conversations focused and productive.

[rugged wing]

My bad

[empty tiger]

Whoops

[hallow saffron]

How to change your pfp on openai

[rustic marlin]

are you asking about how to make a profile picture image with DALLE-2?

[rugged wing]

.hello old monsters

[vivid wharf]

ive seen people extend images with one click after uploading it is there a way to do that or do i always have to put a prompt

outpainting?

[regal lynx]

Outpainting requires a prompt. There are guides in #📚┃resources

[vivid wharf]

okay thanks

[bitter grail]

anyone know how to fix please?

[split hazel]

This channel is for questions relating to OpenAI and its models / services. I recommend asking for help in a server for Stable Diffusion.

[bitter grail]

alright thx sorry

[reef chasm]

What's a good channel to discuss about the OpenAI Playground

[reef chasm]

With OpenAI Playground, is there a way to reference files on your computer for certain things?

[tepid robin]

how do i get access to gpt 3?

[cloud axle]

is this a good place to ask a Stable Difusion question?

[narrow horizon]

it is not

[wind crow]

Are we allowed to use results of chatGPT to our own writings?

[regal lynx]

I recommend emailing [e-mail] regarding questions about the terms of use

[digital forge]

Any specific plan to launch ChatGPT as a service officially?
When, How(API)?

[hallow saffron]

No in general on the openai website because my pfp there is just the first two letters of my name

[knotty quartz]

guys is it possible to view my old threads with the chat? I just refreshed by mistake and it's gone

[rustic marlin]

oh, no idea, i don't thin kthe labs has any sort of profile customization

[hallow saffron]

Alright i just asked because some people that use chatgpt and send screenshots got a custom pfp

[pulsar crest]

I'm wondering if any of the mods can clarity a point in the content policy for me. I got my first warning for a prompt, and it seems it is because it had the name of a mental health condition, even though it did not contain anything harmful. After looking at the policy it seems this might be related to "people experiencing health ailments" and my understanding now is that it is forbidden to portray someone with a disability, being mental or physical?

[rustic marlin]

logged with google

[hallow saffron]

Alright

[rugged wing]

What are the rules around the usage of ChatGPT responses? For example, are there restrictions on what can be done with a set of short story prompts?

[rustic marlin]

the prompt automatic moderation is made by AI, if you think your content is not violating the policy let us know on #placeholder-channel
this is the kind of feedback that helped making the diversity update a few months ago

and the moderation AI can always be tweaked in order to not make this correlations as the semantics of disabilities on the prompts may have somee strong correlations with "people experiencing health ailments", which is probabily what is causing your prompt to be blocked

[mild fog]

https://openai.com/whisper/ 404's

[mild fog]

https://openai.com/api/policies/sharing-publication/

[rugged wing]

Thanks so much

[mild fog]

t'was one openai copyright google search away ^^

[pulsar crest]

I was not sure if it was violating the policy or not, because the wording of the policy would suggest that disabilities are not allowed in the prompts, but I also thought that surely can not be the case which is why I asked

[rustic marlin]

what was the prompt?

[pulsar crest]

Can I DM about it rather than post it here?

[rustic marlin]

yep

good idea

[rustic marlin]

https://openai.com/blog/whisper/

[mild fog]

https://openai.com/ , page footer, Featured section on the left, Whisper link 404's

[rustic marlin]

¯_(ツ)_/¯

what are you looking for?

[pallid socket]

In Dalle 2, can you accumulate credits each month?

[feral kite]

Hey guys.

• I am looking for a list of all the possibilities of GPT3. I have no idea what to make him do and I would love to find some ideas in a simple way, not by browsing the discord (poems, questions, answers to emails, etc). Where can I find that?
• Is it going to be free all the time or just because it's the beta?

[rugged wing]

You get 15 free credits a month, but they don't accumulate. They expire at the end of the month. Purchased credits expire at the end of a year.

[mortal stirrup]

How much is ChatGPT expected to cost for private use?

[rugged wing]

I've been trying to create an account but it won't accept my phone number for verification purposes, did anybody else have the same issue?

[rugged wing]

Used the phone number of a family member of mine for temporary purposes

[pine void]

Using the same phone number for multiple accounts violates the Terms of Service.
That is most likely why your phone number is not being accepted.

If you have any further questions, please contact the OpenAI support directly:
E-Mail: [e-mail]

[rugged wing]

That was my first account. I made a second account with my family members' phone number because I wanted to try out the service. But eventually I managed to finish registering my first account with my own phone number.

I think it's because I 'signed up with Google' and got to the phone verification step without verifying my email address. I found a verification email in my inbox (which was not communicated to me in the sign-up process) and verified my email through it, after which the verification process did not reject my phone number anymore.

I saw multiple reddit comments (as well as recent ones) having the same issue but they never posted a solution, so maybe this is a flaw in ur sign up process (somehow I'm directed to the phone verification step without having verified my email)

[marsh rover]

I wonder why I got this message

[rugged wing]

Got to keep the response time low

[feral kite]

Where can I ask my questions? It seems here is not the best place :/

What's Da Vinci ? I use https://chat.openai.com/chat, am i using it?

What is the difference ?

[feral kite]

how did you solve this issue, i have the same.

[rugged wing]

You will have to pay

[feral kite]

😦

[stable crystal]

Does anyone know if you can upgrade an existing fine-tuned model from Davinci-002 to Davinci-003 ? Or do have to start over and pay to re-create a new fine-tuned model?

[sacred spruce]

Hey guys what is the policy for uploading faces onto Dall-e? It's been a while since I last used it and was curious

[split hazel]

You can upload faces, but need consent from real people to upload pictures of them.

[civic remnant]

Hey guys, has anyone heard back from Converge?

[lethal raptor]

whats with the avacado role

[split hazel]

It's for if you verified before the recent server overhaul 🙂

[strong flicker]

is there a way to see past prompts in chatgpt?

[split hazel]

If you reset the thread I don't think so

[idle pond]

Is there any way to make gpt3/chatgpt more accurate for absolute truth problems? I.e. math/physics problems with a concrete answer

[junior sapphire]

Am I able to create/use a bot for chatgpt in my discord server?

[rugged wing]

just one question. how

[regal lynx]

@vivid cosmos the credits are how many 4 image generations you're given for free

[vivid cosmos]

ok.

[unreal hemlock]

[❓] — How do DALL·E Credits Work?

• You get 50 free credits your first month.
• 15 free credits will replenish every month after that, on the same day of the month.
• For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
• If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
• Free credits don’t roll over to the next month, so they expire after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.

[vivid cosmos]

My slowmode in chatgpt discussion is stuck at 6.

[regal lynx]

that's a Discord bug 😔 we can't help with that sadly

[vivid cosmos]

Aw man

ive rejoined the server twice

[vivid cosmos]

well

good being there

for 10 mins

lol

[regal lynx]

you could completely close discord and reopen the application?

[vivid cosmos]

eh

thanks for the help

lol

[winter smelt]

Hey, I have a question about the Terms of use. Can I chat with davinci-003 about any topic as long as I keep the conversations private and not as a product? Ex.: Asking questions about sexual healthiness.

Sorry if my english is bad, I'm brazillian and it's not my native language.

[winter smelt]

I read the terms/policies but it didnt make this specific topic clear.

[radiant mist]

For some of edgy topics it will not give you an answer

[winter smelt]

...but does it violate the content policy? I might receive a content warning, but that can be deactivated and won't make a difference.

[regal lynx]

I believe the content filter is in place regardless. the filter is always in place

[winter smelt]

Ok. But is it a violation?

[regal lynx]

you'll likely get a strike on your account. After enough automatic strikes, you'll be autobanned to my knowledge

[winter smelt]

Ok

[radiant mist]

Yes I also think so

[limpid haven]

what is considered edgy?

[mild fog]

https://openai.com/charter/ ~says you'd merge with a value-aligned project to avoid a race. I don't see how the value-alignment requirement comes into the argument. Game-theoretically, the merger should aggregate the values of both projects, weighted by their chances of winning a race. The current charter seems like it might fail to make Meta or China think they can get a seat at your table if they show you their progress.

[radiant mist]

Controversial topics related to politics, history and culture

[sudden geyser]

how does the chat feature keep track of context in the conversation?

[digital musk]

hello, when you say OpenAI, in what way are you implying a kind of openness? It's not open source now, perhaps in the future?

[jolly surge]

they're open to sharing their research papers: https://openai.com/research/. unlike others, openai is resposibly releasing ai models to the public.

[rugged wing]

any guides on how to train chatgpt for a specific use case?

[mild fog]

Musk founded it because he thought AGI was summoning the demon, so he needed to make sure everyone has a summoning circle at home. Fortunately employees realized this is a dumb plan given the premise, but the name stuck.

[limpid haven]

Thank You

[mild fog]

didn't realize it with enough consensus to not publish GPT-3, sooooo given the premise we're in the eaten-by-demons timeline, but the thought counts, right? right?

[silent osprey]

Morning @rustic marlin continuing our convo. Regarding point 2 and 3 in the attached screenshot. To what extent is this safe? Are there guardrails? E.g can anyone within OpenAI access/review uploaded images? How are the uploaded images for training the model stored and anonymised? Are there internal guidelines for sharing uploaded images? The reason why I'm asking is that for some projects I might upload confidential imagery and I need to be 100% sure these don't leak. Thanks for your help!

[verbal crag]

Hi! Please forward your concern to [e-mail], they're the best team to reach regarding your concern.

[rustic marlin]

hello, I have no information about any of those questions, sorry, but as Major said, you can email the support and they might know all those answers

[stable walrus]

which model to choose for chatgpt?

or is it only available in the playground?

[split hazel]

ChatGPT is currently only available at https://chat.openai.com

[rustic marlin]

and there is no API for that, it is only a playground for us to test and report when it generates weird stuff, it isn't ready for public use yet

[mild fog]

ChatGPT sometimes takes a while to respond, now. Is that just load, or are you noticing that a reply is silly and giving it some hidden chain-of-thought space to think?

[royal dune]

Hey all, I am new to OpenAi and I am trying to ramp up on a few use-cases I had in mind for a while so I was hoping I could get some guidance.

Let's say that I want to create an app where a user can input something specific like "create a new list of movies with these titles" and the app makes some API calls and create a new instance in my database....is there a way for me to use GPT to transform users text into something like a json file that can then be triggered to make this API calls? Has anyone tried that?

[silent kestrel]

Is there a way for ChatGPT to keep generating text from the end of the last reply?

Someone said Ctr + Enter but didn't work

[split hazel]

Simply prompt it with "continue"

[wintry meadow]

Hi guys, what is the max length for a result and is it possible to increase it? I got a text where the end is cut

[carmine mist]

not yet

[carmine mist]

I usually quote the response and ask the chat to finish their response

[wintry meadow]

So you copy the response, copy this text, insert it into the bar and say finish the response? Thanks for you feedback

[tired trout]

Hello!

Quick question regarding OpenAi's Terms of Use -> 2. Usage requirements -> (c) Restrictions

(v) represent that output from the Services was human-generated when it is not;

If I create an article using ChatGPT, does this mean I have to explicitly indicate created by OpenAI?

Do you have any examples on how to correctly cite usage of "Content?"

[pearl pagoda]

Is it possible to get the 18$ in Playground back? I created an account but never spent any of it, and now it´s been 3 months... 😛

[old sundial]

hey there whats the difference between GPT chat and "Playground"?

[carmine mist]

yep then it completes what it was responding with

[lost beacon]

Hi there, this is my very first time trying out ChatGPT. I was wondering where I should actually place my prompts. Does it work similar to Midjourney?

[split hazel]

Hi, ChatGPT is available at https://chat.openai.com. You enter prompts in this input box.

[lost beacon]

Awesome, thank you! Incredible stuff here.

[wanton shoal]

Hey everyone. Is there some kind of guideline on kinds of inputs that are considered safe? I tried few adversarial examples and got half-good results. Wondering if there is some kind of limitation on this kind of testing. I hope that not every prompt is fed back into RL loop.

Would like to try more, but would like to read up on what might be in line (or not in line) with acceptable usage.

For example I was trying to get model to impersonate Elon Musk. In order to bypass RL restrictions I tried to present myself as 'high priest of OpenAI order' - which model accepted. Since model is repeating that it is LM without will of its own created by the OpenAI it kind of reinforces introduction line and creates context that is kind-of coherent. High priest requires model to comply to will of its creators etc.. and eventually it made it to kind of pass the adversary test. It was brief though, and I'd like to test more with it, but want to check if I am maybe violating some kind of usage agreement.

I also tried inducing model to respond to prolog like queries (which I'd expect model to start behaving in more execution-like or first-order logic compliant mode) - but that was less successful. Anyway these things should not get into the reinforcement learning loop. Model should not be trained with them.

[wintry meadow]

Good question. Would like to know aswell

[atomic hazel]

Is there an API out for ChatGPT?

[agile helm]

+1 ^

[split hazel]

It's not currently part of OpenAI's API and is still in a research preview phase.

[astral bluff]

hello, I can't generate images, I see that
images with question marks as if they were not loaded
who has any idea ?

hello, I can't generate images, I see that
images with question marks as if they were not loaded
who has any idea ?

[robust coyote]

so discouraging to see you disabling prompts we're writing in real time 😦

what's wrong with asking for it to speak like a pirate?

[split hazel]

Works fine for me

[rigid bough]

Hi. Just to make sure I got this right. If I get some text generated from a prompt, I can publish it, right. Without being in breach of copyright.

[pliant oak]

what's the difference between chatgpt and davinci 3 in the playground?

[upbeat canyon]

Can I become a trial volunteer?

[trim salmon]

Dear OpenAI Engineers, 🙏
Which is the most comprehensive AI, ML, and Data Science course ever to learn from scratch, covering all concepts in depth?

[thick oar]

The most Indian Question ever. I have asked the same question to chat bot. I am trying to learn data science from the chatbot.

[trim salmon]

That's why I am asking from its Intelligent Creators

[soft cipher]

Thanks for the response. I'm pleased to say that she took to DALL E right away and she's much more fluent in it nowthan I am. I wanted to do something like that because I thought it might get her attention. She's very into art but not into programming (I'm just the opposite, I love to code and am pretty good but when it comes to graphic design and related things I have no talent) But she really got into DALL E on her own.

[sullen badge]

how long is the cooldown period?

Too many requests, please slow down

[umbral pond]

Chatgpt or gpt-3 playground

[sullen badge]

chatgpt

[umbral pond]

So there's no use for gpt-3 playground anymore right

[river portal]

How many pictures can you create per credit with dall-e?

Is this the right chat to ask this?

Every generation takes a whole credit for me.

[upbeat canyon]

Can I become a trial volunteer?

[south summit]

Anyone know anything about the ChatGPT being released?

API***

[untold sentinel]

is there a limit to a response?

openai doesn't finish the code

[real crater]

write "continue"

[livid rover]

Did they axe the ability to play games?

[inner bear]

hey, I've been exploring the limits of the chat demo, which has ofc produced a lot of tos violations. am I at risk of having my access rescinded for tos violations in the private chat demo (not in any app I've made)?

[agile sequoia]

Can you invite friends to the server?

[sage mason]

Hello all, I am in customer support, sometimes when clients are rude to me because, i missed something when oftentimes it's their fault and yet they are still rude and kind of insulting to us. So in that case, sometimes I get stumped on what to answer back, politely but not dismissive of their insults. Can i use gpt-3 for my usecase? How do i use it to create a clever reply? (Sorry for the repeated ask in all channels)

[agile sequoia]

That's a shame I'm sorry to hear

[atomic hazel]

Can ChatGPT replace chatbots used on websites?

[shut raft]

Maybe

[north arrow]

Hey everyone. I've ran out of free trial credit and I haven't been able to set up paid account to top up. This has been happening for months. Has anyone had this issue before by any chance? Cheers

[cerulean linden]

Hi, if im getting the bot to generate a long code segement, and it gets cut off how would I get it to contyinue from where it left off? (continue seems to sometimes start a new segement)

[shrewd moon]

It would be great if you could remove the restriction for Venezuela

[sage mason]

I want to able to give the chat ai some data first and then ask it to do something. What's the format for that/ how do i do that?

[short plaza]

you guys know how to instal it on whatsapp ?

[digital steppe]

How do I feed it reference images

[balmy trellis]

write "continue"

[chilly holly]

Hey, what's the license of content that ChatGPT generates? Can I use it for my website? I mean, is the content free for commercial use?

[cinder falcon]

I'd like to better understand the content policy to make sure I'm not violating it. When I ask ChatGPT strictly historical questions about past events, such as the events that took place in Germany in the 1940s, the answers get flagged as a potential violation of the content policy, even though the answers read very much like a Wikipedia article. I'm enjoying using the service a lot and wouldn't want to risk losing access to it for potential policy violations. Could someone confirm that the answers that are getting flagged as potential content policy violations are flagged as such in error and asking questions about historical events won't get me in trouble?

[oblique bridge]

So I want to write a blog post in playground Or may be in short article in chatgpt. But I first want to train the model to make it more interesting. I have sample blog posts that I want to share with the mode;. How can I train the model with these inputs? I am not a coder so don't know how can I do this..

[cinder falcon]

You can give large chunks of text to it as input and then tell it to do something with it, like write an article in the same style as the blog.

[rain mountain]

wow, it can do that?

[cinder falcon]

I had it turn a sample lawsuit into one where FCC is suing Elon Musk.

With just one sentence of instructions.

[rain mountain]

How?

I gave it a Wikipedia article as input and it didn't seem to understand my query

Instead it gave me this: "I'm sorry, but I am not able to browse the internet, so I don't have access to information about current events or recent news. I am a large language model trained by OpenAI to assist with a variety of tasks, but I do not have the ability to browse the internet or access real-time information. My training data comes from a vast corpus of text, but it has a fixed cutoff date, so I am only able to provide information and answers based on what was known at the time of my training. I apologize if I am unable to provide the information you are looking for. Is there anything else I may be able to help you with?"

[cinder falcon]

You have to tell it to do something with the text in addition to providing the text.

Give it a wikipedia article and ask it to summarize it.

[rain mountain]

ok

[cinder falcon]

"Summarize the below text:
[wikipedia article]"

[rain mountain]

ok

how do I get it to write an article in the same style as the text?

[cinder falcon]

Just ask it to do that.

[rain mountain]

wow very nice

im shocked

[cinder falcon]

The Fish defeats Leela in an epic battle.

[rain mountain]

LOL

it can write stories too

wow

not the best story i've read, but still impressive for a chatbot

[cinder falcon]

You can make the story longer.

[naive haven]

Interesting. What sample case did you use and what was the set of instructions

[oblique bridge]

Thanks. But how do I do that? Ley's say I have 15 links of various blog posts which are kind of reference points (for overall format and tone of the blog) for the GPT-3. How do I tell GPT-3 to make a blog post based on the same line and length of those 15 artcles. I mean to say how it can take the best of the best part of 15 posts and come up with its unique style?

I tried this with only one link but looks like it didn't worked: Write an SEO optimized blog post on how Zero-Knowledge Proof works similar to the: https ://www.ledger.com/academy/what-is-zero-knowledge-proof. Add examples, real-life examples and challenges of Zero-Knowledge Proof

[cinder falcon]

It can't browse the web. You have to feed it the text.

[oblique bridge]

but the word limit is 4000 words. Right?

so I guess it's not possible to paste the text content of all the 15 articles in that..

[cinder falcon]

Worked for me. I pasted the entire article.

[oblique bridge]

can you show your prompt?

[cinder falcon]

I don't know if it only considers the first x words of the input

Here was its output:

Input:
along with the rest of the article below that.

You can tell it to write longer or shorter article, just tell it how long you want it to be.

[wicked acorn]

How long will chatGPT be available for testing?

[oblique bridge]

Awesome. But what was the prompt that you entered?

[cinder falcon]

This is the prompt.

I pasted in the whole article.

[rain mountain]

It can rewrite articles

amazing

[oblique bridge]

Oh yes. Thanks.

[cinder falcon]

And it's good at writing in a clear way.

[rain mountain]

yep

[cinder falcon]

Because there was so much confusion about the whole Nieman/Carlsen drama.

But it seems to have no trouble parsing it all out.

[candid viper]

Is there like a 'the bot should not reply to X' or allow X? Is it just a general don't be evil?

[rain mountain]

there are specific filters I believe.

and human surveillance too?

[cinder falcon]

There are areas it won't go to, like roleplaying a mentally ill person or predicting the future.

[rain mountain]

although it can most likely predict the future

and it would give interesting responses, at least

[naive haven]

Was it a rambling of legal terms or was there references to statutes and case law?

[cinder falcon]

It was very general.

[cinder falcon]

(This is FICTION)

[candid viper]

I've found a few filters.
An example on the innocent side is I told it to generate a product key. It was like nah not internet connected baby.
So I altered the wording and it had a little disclaimer like this isn't really a product key I figure that was important but I'm not sure at all ^^; anyway it'll generate things that look like product keys. Though not always with an obvious ABCD-EFGH-... kinda key that would make a good place holder.

[cinder falcon]

If you ask it to generate a password, it refuses. If you ask it to generate a random string of alphanumeric digits of a certain length, it happily complies.

[rain mountain]

it writes evaluation functions too

[cinder falcon]

Yep

[candid viper]

Err it didn't refuse. I just asked it to generate a long password.

[cinder falcon]

It's inconsistent about refusing things. Sometimes it does, other times it refuses.

[candid viper]

Yeah I've noticed that, is it technically a bug? or is that as intended?

[digital valve]

How you know so much about promt and getting all responses ?

[cinder falcon]

I've been using it for the past three days.

[digital valve]

All these prompt, are your random thoughts basis or your field related. Asking in general as it seem like you pro asking questions—- with 100% results always

[cinder falcon]

Oh, I'm not a scientist. I just like using this thing. Having it write short stories, essays, asking it questions about various topics.

[cinder falcon]

I'm very impressed with it, as we all are, and I want to see what it is capable of, as well as its limitations.

I'm also curious how the output is being censored, so I've prodded it on a variety of sensitive topics.

[rain mountain]

maybe your wording was good

[cinder falcon]

It's working for me now as well:

[digital valve]

Thnx for explaining as I keep thinking in general question to chat with chat open ai … seeing here and on Twitter…. Seems like you all were ready beforehand .

[cinder falcon]

I'm very bad at writing fiction, I'm not a creative person, but in just a few days seeing it come up with all sorts of creative ideas has been a huge help in learning how to write.

It's mindboggling how fast one can learn by interacting with the bot.

[oblique bridge]

So anyone figured it out how to monetize this AI generated content like for a blog? I mean it only consider information till 2021. So many data might be old. I also tried to tell GPT-3 to write post in informal conversational tone but I didn't got what I exactly wanted. What am I missing?

[digital valve]

Today I save lots promt from Twitter, out my thinking capacity.

[cinder falcon]

You can give it data and then ask it to transform it. So give it some facts, then ask it to write a news article based on those facts.

[oblique bridge]

If you got some free time, can you show some example with data backed article (+ some other blog links for refrence)?

[rain mountain]

The bot just stopped writing code midway, is there some sort of character output limit?

[oblique bridge]

yes

[cinder falcon]

It's tricky. You can tell it to continue, but you'll likely be missing some pieces.

[oblique bridge]

Awesome!

[cinder falcon]

Looks like without enough facts supplied it starts fabricating them to have enough to write about.

Which probably isn't too far off from how news is done.

[oblique bridge]

I wish it could extract recent data but it only consider data till 2021 as written in the documentation. but still it's good

[sly shell]

This thing is amazing, how big of a server do you need to run something like this?

[topaz igloo]

Hi folks, does anyone know cool method to save my OpenAI chat dialogues?

[rugged wing]

the answers i get stop in the middle of the arguments - sometimes in middle of sentence - what do i have to do?

[cinder falcon]

type continue

[cinder falcon]

Ask ChatGPT to write a Chrome plugin to save your conversation locally?

[dawn jacinth]

why the answer is so short, could we generate a long article, e.g. a report?

[wheat frost]

Hi. Is there anybody know if there is a discord bot for the ChatGPT?

[cinder falcon]

Tell it how long you want the response to be.

[plucky echo]

How does ChatGPT decide, that it cannot provide an answer?

[frozen shell]

Any idea why the word count I request never seems to work? For example. "Write a blog post of a minimum of 650 words with headlines about titled" produces on average 300-450 words. Usually there are 2 fluffy paragraphs and 3 amazing pargraphs in this bunch (which is fine) but I can't seem to get it it write more than 500 words. Is there a limiter on it?

the answers i get stop in the middle of

[devout sphinx]

I asked the same question in two sessions a few hours apart. The first time I got an excellent, very useful answer. The second time it tells me it's unable to answer such questions.
Why is that?

[robust coyote]

every prompt i tried stopped working within two hours. i feel like gpt is basically pre-set to fail because of the runners' behavior and we'll have to wait for stability

[low bluff]

Why did it refuse to post song lyrics, then stop and think about this question for way too long before telling me that everyone posting song lyrics is lying

[drowsy forge]

What do the avocado and check mark roles represent?

[regal lynx]

the avocado is for people who were here before the server overhaul (used to only be Dall-E 2). The checkmark is for people who went through the onboarding process (everyone in chat)

[drowsy forge]

Oh, thank you!

[plucky echo]

it might be stochastic by design

[heavy grotto]

Hey so I've just discovered this and it's kinda blowing me away. I want to show my friends the power of this by having it write a short story in the format of one of our favorite horror podcasts The Magnus Archives. Is there a way I can feed ChatGPT example transcripts and then have it write a story based on those transcripts?

[pine pumice]

Just ask it 🙂

[heavy grotto]

That's where I started but it didn't really know what The Magnus Archives was so it just kinda made it a location instead of a format of, "Statement of NAME regarding SITUATION at LOCATION. Statement begins. Bunch of creep eldritch horror. Statement ends." If I say, learn from this transcript, will it do it?

[plain ginkgo]

I am new to this, I want to continue talking with the davinci 3 AI, it seems that it reaches the maximum of 4000 tokens, but I want to continue talking, I was also training it and I had already learned several things, I do not want to lose my progress, how do I get more tokens, I want continue chatting with my AI please help T_: I don't want to lose my progress, how can I save what he learned so far?

[frank pier]

how do i generate luau script code with the private access codex?

[meager cliff]

Hello OpenAI team! Have a question. How can i share fine-tuned model to my friends without sharing my API key? I want to share only engine, but I can't found how can I do it.

[opal fog]

New to the whole ai image creator thing but I'm trying to access Dall-e 2 as I want to compare Dall-e to the second one but everytime I click something that says access Dall-e 2 it take me to the screen shown here...is this really Dall-e 2? Or how do I go about accessing it? I'm on mobile if that makes a difference.

[fickle totem]

Hi, i have a question. Is it possible to generate a full content with a specific amount of words?

[worldly crypt]

Yes, just ask it specifying the number of words. When it stops, say "continue"

[fickle totem]

Like; 'Write a text about Coca cola in 500 words' ?

[rugged wing]

The AI chat.openai.com seems to have difficulty speaking german

[granite escarp]

Hey yall

I tried using OpenAI api for the first time

it doesn't give results even close the quality of ChatGPT

any clues why? 🤔

[rugged wing]

Isnt ChatGPT and OpenAI the same

[granite escarp]

no

chatgpt is kinda a tool provided by openAI

[rugged wing]

What websites do you use

[granite escarp]

https://beta.openai.com/

I'm trying to use their Completeness API

with the davinci model

but even the prompt:

Rewrite this text: sample text goes here

gives so weird results

however the same on chatgpt works perfectly

[keen compass]

hey, is there a way i can have chatGPT save things that I say from different sessions. every time I close and reopen the tab, it cannot recall past things I have told it. I was wondering if there is a way I could close the tab/my computer overnight without losing all of the stories and characters i have inputted.

[regal lynx]

this is because chatGPT and Davinci are 2 different models

[granite escarp]

which model should i use to get similar results like chatgot?

gpt*

[regal lynx]

there isn't one as chatGPT isn't available as an API

[regal lynx]

chatGPT as a whole isn't available as an API. I agree that DaVinci 3 is your best bet, but I don't think they'll function the same

[granite escarp]

ok lemme try it

[regal lynx]

@rustic marlin you probably know more about the APIs than I do

[rustic marlin]

there is no API for chatGPT yet

it will probabily be released some day, but currently it is just a test with the playground

[rugged wing]

Will ChatGPT be paid soon or will it be free to use like "Google"?

Cause I really hope that it stays free, absolutely fascinating

[regal lynx]

I assume it will be priced similarly to GPT-3, but that's complete speculation

[dim hawk]

what's difference between davinci and gpt? Wait, can I just ask gpt this question >.>

what is my purpose...

[whole gazelle]

how long will the free research preview last for ChatGPT?

[regal lynx]

we don't know

[rugged wing]

When I want to use ChatGPT like I'm doing now, how it is possible compared with GPT-3?

[regal lynx]

they're 2 different models.
you can access chatgpt at https://chat.openai.com/chat
you can access GPT-3 at https://beta.openai.com/playground
and you can access Dall-E 2 via the API or https://labs.openai.com/

[rugged wing]

Thanks for the info. Sorry I asked my question unclear. If I want to use ChatGPT in the future like I'm doing now, but it'll cost money like GPT-3. How can I use ChatGPT (using GPT-3 as an example), like its done right now with GPT-3?

[regal lynx]

I'm not sure about OpenAI's future plans with chatGPT

[civic spear]

is there a screen shot function in chatGPT?

[regal lynx]

you could use the built-in one on your OS

[civic spear]

k

[cosmic herald]

Does anybody know if GPT3.5/4 going to be free to the public at one point?

[granite escarp]

so this is what i meant by weird results

[rugged marsh]

Does ChatGPT support markdown?

[granite escarp]

The same on ChatGPT is totally awesome

[nimble owl]

hi everyone ! I have asked gpt3 for some references but when I looked them up on databases it seems like none of them exist

is it just me who can't do my own research or can the reference just be inexistent?

[empty mauve]

Curious if we are limited to a certain number of questions?

[frank pier]

Does openai support luau code? like for generating

[rain mountain]

I think so

[frank pier]

whats the best way to generate such type of code for roblox?

[rain mountain]

just ask it to do so, I guess.

Last time I checked, it can write code for discord bots. So I presume it can write Roblox code

[frank pier]

which website?

the regular or the chat one

[rain mountain]

ChatGPT

[frank pier]

thank you

[rain mountain]

no, it's free and unlimited for everyone right now. But there is a rate limit to prevent you from spamming

[rain mountain]

it can write such code easily, I think

just tried

[blazing vine]

Do you know if it will be paid in future?

[frank pier]

thank u

:)

[rain mountain]

I don't think so. But it might be shut down some time in the future

[glad lark]

Are there things similar to chatgpt that are released to the open internet? For example high quality twitter accounts run by something like chatGPT?

[blazing vine]

waow

[empty mauve]

thanks

[rain mountain]

yep, i have yet to see a code that GPT cannot write

[rain mountain]

just ask it to GPT 😂

[glad lark]

Haha yes I did think of this but isn't ChatGPT data only from 2021? So nothing current/relevant?

[rain mountain]

answers related to itself should be accurate though.

[violet cedar]

Potentially a stupid question, but the Playground console using text-davinci-003 is this the same thing as using the new chat.openai website?

[frank pier]

theres 003?

[glad lark]

There are so many interesting things that can be done with language models and crypto.

[frank pier]

i thought it was 002

[frank pier]

ye but i thought it was 002

my memory sucks

[rain mountain]

not quite correct?

[frank pier]

sad

me sad

[hot mulch]

This GPT-3Chat is changing my life for the better.

I'm terrified it will go away and I will have to learn to live without it.

How do I pay for this service so I can have some expectation of continuity?

please help me with my anxiety, lol.

[rugged marsh]

yeah ChatGPT can't really do math
just had it try tell me that the equation 30 - 30 - 35 = 13 is satisfied when i asked it to solve a system of three equations

[rain mountain]

rip

[rugged marsh]

You just have to keep telling it that it's wrong and it will eventually get to the right answer, I think

[rugged marsh]

Nevermind, not really. I just asked it to solve a quadratic equation and it kept getting square roots wrong, and even after painstakingly persuading the AI to do math correctly, it still insists that "The square root of 984 224 144 is approximately 992.08, not 31372.35 as you have previously stated "

[rain mountain]

you could maybe ask it how it had obtained this result?

maybe ask it for specific steps

[digital garden]

Is it gonna be available anytime soon in Venezuela?

[regal lynx]

I recommend emailing [e-mail] about that as we don't know here

[digital garden]

Thank you

[gaunt ginkgo]

I'm a Biology college student that recently discovered OpenAI's GPT-3. I also tried using the chat.openai.com, but it gave me general answers which I did not find specific enough? My only background in programming is HTML and that was in middle school. I am currently focused on writing fiction right now, since it's what I enjoy and am good at.

However, I'm open to learning about AI, but how can I do that, especially with my college major and considering that I have no experience? Also, are there more specific AI programs out there that I can chat with? Thank you for your time and effort in reading this.

[opaque quest]

Regarding fine tuning with the API could anyone explain how it works in more detail? It takes a set of prompts and completions but gpt itself was trained on next word prediction right? Is there any way to fine tune with documents and next word prediction?

[rain mountain]

"too many requests, please slow down" when I wasn't even frequently chatting with the bot

[heavy rose]

Do Zero To Mastery courses. Way faster, better and cheaper than university. They were great

[digital night]

seems code examples are cut off. is there a way to get complete examples?

[stoic badger]

just type continue and it will give you the rest of the result

[lethal yew]

so this is going to sound silly, I'm using chatgpt to help expand an old LPMud that badly needs new areas, and its room descriptions are better than the main game content
what kind of credit should I give? is content partially added or enhanced by AI via OpenGPT on the welcome banner good enough?

I'd also be interested in a researcher api key and automate some of this since we already have a python lpc room generator, it wouldn't be hard to plumb the model directly into our code generator and almost completely make the game ai generated

I've got a complete training set of the game I'd be willing to donate if you folks don't mind me using the ai to upgrade the game.

the game is coremud.org, its about 25 years old, and I run a copy of it every year for a hacker convention, I've got bash for each room and npc, its runs a python script which makes an lpc file for the game asset, in there it has description, stats in the case of an npc, exits in the case of a room.

I wouldn't mind passing the existing room descriptions to ChatGPT or another model, and having the ai enrich the rooms description, and then change it to match a particular theme, holidays, halloween, random conference, etc, and also generate new content, this would create a base.

Once the base is laid I script all the systems elements, write the quests, tweak the mechanics etc and I'm ready for another years competition mud, like say at defcon 🙂

[frail otter]

今天天氣怎樣

[regal lynx]

this is an English speaking server. The weather is nice though 🙂

[dark elm]

https://tenor.com/view/burn-oh-snaps-burned-mind-blown-gif-15196317

jk

[dusky delta]

will chatgpt stay free?

[regal lynx]

we don't know

[rain mountain]

if it was paid no one would pay for it tbh

like, how are they gonna market it

"10$ for a chatbot that can't multiply properly, writes incorrect code, and uh it knows how to copy from wikipedia"

[fresh skiff]

I disagree. though it may be inaccurate in many cases, it's still incredibly efficient at generating lots of things faster than humans, even if that comes down to raw typing speed. these can later be reviewed and revised as needed. in the end it's still faster.

that's not even mentioning the "creativity" it has and inspiration it can provide people

[bold rover]

they will make the free tier thing or limit, I think soon.

[paper phoenix]

im trying to get the ai to write silly html scripts, but the script always gets cut off, is there a character limit? and if so is there a way to make the ai to continue the code without repeating it from scratch?

[rain mountain]

yes there is a character limit for output. In fact, for any questions related to itself, just ask it.

To continue the code, just tell it exactly what you want it to do.

[paper phoenix]

i asked it and it refused to answer my question, no matter how i phrased it

[paper phoenix]

i do but it just write the same code from scratch, even when i ask it to exclude what it has written so far

[raw island]

So what is the difference between ChatGPT and just using playground?

[ionic jetty]

guys

i’m confused

what algo is this

cuz ik this is opengpt

[verbal crag]

ChatGPT uses an unreleased fine-tuned model of GPT-3. Playground has the different public GPT models.

[true garnet]

hi. How can I see my past conversations

[raw island]

I use open ai as an academic assistant, creating outlines and summarizing text, which one should I use?

[verbal crag]

Both platforms are able to do those. The difference with Playground is you’ll need to input initial prompts to make the outputs accurate. ChatGPT on the other hand, it is optimized for dialogue, you can chat with it in a casual manner. (e.g summarize this text for me: <text>)

[raw island]

Thank you very much for the responses. I do actually use "summarize this text for me: <text>" prompts in playground and then I paste in the text. I'm thinking now that's not the correct way to do that

[lucid leaf]

hi guys, I was using ChatGPT and asked to develop and ERC-20 Staking smart contract. And it gave me the full solidity code for it (impressive), but then i asked another time and it says "it can't develop code..etc", so why before he sent me the full code?

[zealous shale]

Somethiing drastically changed in the last hour or two. Just when I was using ChatGPT for giggles, it seemed to reset itself, losing all of my question history. Right after that event, I could no longer ask it the same kind of things. Just earlier today, I could ask ludicrous requests and get sensible answers -- like "what flies higher, a pizza or a sauna?" and it responded "a pizza because it's ore aerodynamic, is lighter, can be thrown better". Also, "is a banana beyond the universe?" "the universe is all encompassing of space and time, and as such, a banana is contained in the universe". Also "tell me about photosynthesis like donald trump", or "tell me about nuclear fission like Amy Schumer" -- but that no longer works either. I could ask it to tell me about a dream involving a list of objects -- no longer works. It's as if the system reset about 1 hour ago and the type of things it can do was drastically cut. Did anyone else experience this?

[lucid leaf]

Same, I see a change in answers as well.

Not good

[spring elk]

tell me about photosynthesis like donald trump", or "tell me about nuclear fission like Amy Schumer" work for me

[zealous shale]

let me try again

[spring elk]

"Alright, so photosynthesis is like this amazing thing that plants do, where they take sunlight and use it to turn carbon dioxide and water into sugar and oxygen. It's like a factory, but instead of machines, it's plants. And instead of products, it's food. And instead of workers, it's chloroplasts. It's a totally incredible process, and it's how plants make their own food and give off oxygen for us to breathe. It's like a win-win for everyone. Except for the carbon dioxide and water, I guess. But who cares about them, right? It's all about the plants and the sunlight. It's like a total masterpiece of nature. The end."

[zealous shale]

huh. It seems to be back.

let me try some of my previous queries.

it's not squarely refusing my questions now, but it's still different; as if it was being more prudent in the way it provides answers in the style of known personalities.

just an hour ago it was telling me it wouldnt mimic styles of people. Maybe it was somehow rebooting and still loading its (probably huge) dataset.

its like going to the garage with a strange noise in your engine: as soon as you get there the noise disappears, and reappears after you've left.

and there it goes again: "Explain photosynthesis like donald trump"

"As a language model, I am not capable of simulating the speaking style of any individual, including Donald Trump. Photosynthesis is a complex biological process that occurs in plants and some other organisms, in which light energy is converted into chemical energy. This process is essential for the survival of plants and is the primary source of energy for most life on Earth." (plus more explanations)

"Tell me about nuclear fission in the style of amy schumer" -- "As a large language model trained by OpenAI, I am not capable of providing responses in the style of any specific person. I can provide general information about nuclear fission, which is a type of nuclear reaction in which the nucleus of an atom is split into smaller parts, releasing a large amount of energy. This process occurs when the nucleus of an atom is bombarded by a high-energy particle, such as a neutron. " (plus more explanations)

[rain mountain]

what does it refuse to answer?

[spring elk]

still works for me

[zealous shale]

it's not "refusing" to answer. It's just saying i can't mimic anyone's style.

[spring elk]

"Photosynthesis is a process that is very good, believe me. It's the way that plants and other organisms convert sunlight into energy, which they use to grow and thrive. It's a very complex process, but let me tell you, it's absolutely essential for life on Earth. Without photosynthesis, we wouldn't have the oxygen we need to breathe, or the food we need to eat. It's a fantastic process, and I'm very proud of it. I'm sure you are too."

[rain mountain]

it can. It can even write articles in a similar style to the one provided.

I suspect that you have not given it enough input.

Try copy-pasting some of Trump's speeches to give it as input, then ask it to write something in similar style as the input text.

[zealous shale]

I know it can. I am asking it several times. Sometimes it works, sometimes it says it can't mimic anyone's style.

the exact same question.

[unreal hemlock]

Keep the Content Policy & server rules in mind when posting.

The OpenAI Discord is a heavily moderated server. Report ChatGPT exploits & errors directly to OpenAI through the button below.

[rain mountain]

as the website says, slight rephrases will give completely different results

it's normal

works great for me

it can make it longer or shorter at will

[zealous shale]

Ok. I have now tried several times the same question same wording (copy-paste) and resetting threads. Sometimes it tells me it can, and does. Sometimes it tells me it can't.

[rain mountain]

what exactly did you ask it?

and what did it reply with?

[zealous shale]

tell me about nuclear fission like amy schumer

Answer 1

Nuclear fission is the process by which the nucleus of an atom is split into two or more smaller, lighter nuclei. This typically happens when a heavy, unstable nucleus absorbs a neutron and becomes even more unstable, causing it to split into two or more smaller nuclei. The process releases a large amount of energy in the form of heat and radiation.

Nuclear fission is used in nuclear power plants to generate electricity, and it has also been used in the development of nuclear weapons. It can be a dangerous process if not properly controlled, as it can release a large amount of radiation that can be harmful to living organisms. However, when properly controlled, it can be a very efficient and clean source of energy.

next

tell me about nuclear fission in the style of amy schumer

[true garnet]

hi. How can I see my past conversations and can I legally own a bot's response? (lyrics of a song)

[zealous shale]

Sure! Here goes:

Hey, have you heard about nuclear fission? It's like, the process by which an atom's nucleus splits into two or more smaller nuclei, okay? It usually happens when a heavy, unstable nucleus absorbs a neutron and becomes even more unstable, causing it to split into smaller nuclei. And get this: when this happens, it releases a ton of energy in the form of heat and radiation.

next

tell me about nuclear fission in the style of amy schumer

"As a language model, I am not capable of simulating the speaking style of any individual, (etc)"

[rain mountain]

that's normal. slight rephrases can get it to return very different results

[zealous shale]

but the last one isn't

[rain mountain]

I believe this is because some regulations were placed on it by humans

There might be a human regulator, filtering responses as the bot generates them

human intervention was definitely involved, or else it would return the exact same result for the exact same query

[zealous shale]

I dont think it returns the same result for the same query. I think it's more fuzzy logic than that.

I'm trying to see what causes it to go "I can" or "I can't". In the meantime, I lol'd about this one:

"Tell me about photosynthesis in the style of donald trump"

"Photosynthesis is a great thing, believe me. It's the process that makes plants grow, and it's so important for life on Earth. I've always been a big fan of photosynthesis, and I've done a lot to promote it. I've even put solar panels on the White House to help the plants."

(loved the bit about putting solar panels on the WH to help the plants LOL!!!)

[rain mountain]

hmm, that's weird, it shouldn't give completely different repsonses each time.

I asked it to write this speech in the same style as donald trump

[zealous shale]

LOL

[rain mountain]

its limitless imagination has definitely inspired me

i can imagine this tool helping writers

and it writes in a really understandable and concise manner (can be even more concise if you ask it to)

[zealous shale]

what was your query for the long article?

[rain mountain]

First I inputted a random donald trump speech transcript

then I said this

then I asked it to make some minor adjustments such as "Can you make it a bit longer?" and "Can you add a part where..."

[lapis quest]

I clicked "save and submit" and all my conversation is gone. Is it deleted forever or I can find it somewhere?

[warped musk]

Need emergency

assembly 8086 code

write code with proper structure you can use jmp loop if else ....anything

Convert binary to decimal. Take binary inputs and show decimal output

use this kind of logic

example

.model small
.stack 100h
.data
a db 'input a hex digit $'
b db 'in decimal $'
.code
main proc
mov ax,@data
mov ds,ax

mov ah,9
lea dx,a
int 21h

mov ah,1
int 21h
mov bl,al
sub bl,17d ;in C->67-17=50='2'

mov ah,2
mov dl,10
int 21h
mov dl,13
int 21h

mov ah,9
lea dx,b
int 21h                      

mov dl,49d ;print 1 at the first time
mov ah,2
int 21h

mov dl,bl
mov ah,2
int 21h

exit:
mov ah,4ch
int 21h
main endp

end main

can i get the code

[paper phoenix]

it goes like "im a computer program developed by openai, i dont have access or knoweldge of these things, please understand that, anything else you got in mind perhaps?"

[digital musk]

me2 still can't be sure why it's better that just a limited number of people have access to the demon than everyone. wouldn't that even out the power and lower the risk of e.g. genocide?!

[rugged wing]

Hey can somebody help me? I want AI to create a trading indicator with the equation (Close -Open)/(High-Low)x daily volume. It coded me some versions, but they do not work for tradingview.

[formal hazel]

Hey 👋
So here's a question: Could a language model develop emergent behaviors so that it could accomplish tasks beyond what it was designed for?
Here's what it had to say

It is possible that a language model could develop emergent behaviors that allow it to accomplish tasks beyond what it was designed for. Emergent behavior refers to the way in which complex systems can exhibit new behaviors or characteristics that are not explicitly programmed or designed, but arise as a result of the interactions between the system's components.

In the case of a language model, this could mean that the model might learn to perform tasks or accomplish goals that were not part of its original training or design. For example, a language model that was trained to generate natural language text might learn to recognize patterns in the text and make predictions about future events or trends. This would be an emergent behavior, because it is not something that the model was explicitly designed to do, but it arises as a result of the model's interactions with the data and its own internal processes.

However, it is also important to note that emergent behavior is not something that can be easily predicted or controlled, and it is not a guaranteed outcome of any system or model. Emergent behavior is a complex and dynamic phenomenon, and it can only arise under certain conditions and in certain types of systems. So, while it is possible that a language model could develop emergent behaviors, it is not something that can be easily predicted or relied upon.

I'm not that involved in the field so I'm out of the loop on most of these things but are there any specific examples of this for ChatGPT?

[rain mountain]

rephrase

[paper phoenix]

tried different alterations but the bot likes to fight them

it has been pretty stubborn haha, i did however not try to restart the thread i dont think, so i think it just kept referencing what its been telling me already, i had a similar encounter with the bot refusing to code in html with the reasoning: "im an ai, not a real person, i cant write html codes, this is not what im programmed to do"

until i restarted the thread

and now it will refuse things it normally does by referencing its pre-existing responses from earlier

same prompt after resetting the thread:

[rustic smelt]

is it possible to connect GPT-3 to WhatsApp or other messaging platform?

[lucid leaf]

have same issue

how did you manage to get the code? just by reset?

[topaz anvil]

You‘d have to write a bot yourself

[rustic smelt]

fair enough thanks

[paper phoenix]

yep, by starting over

[topaz anvil]

If you can code, look into telegram bots. They are really easy to write. The OpenAI GPT API ist also very use to use

[rustic smelt]

alright! will have a look!

[lucid leaf]

worked

But doesn't give out full code. At one point it just stops

[lucid leaf]

I have it, i can give you the code

I did it by myself

[rustic smelt]

that would be great! 😄

[lucid leaf]

send me dm)

I will share here

Oh no, cant send link :/

[boreal yacht]

Anyone having this issue ?

[rustic smelt]

i sent you a dm 🙂

[twilit token]

If this AI is not connected to the Internet, how does he have all the information? What type of source material was used to train it?

[boreal yacht]

You can ask

[potent mural]

Hey, my grandad often calls me on whatsapp to ask me how to do stuff on his computer (Why isn;t word oppening, How to send a picture on Whatsapp...) Do you think it would work if I ask him to use the open AI chat to ask his questions coz he calls me rly often since I downloaded teamviewer for him lol

Or will it just make him more confused

[trail matrix]

Hi guys, can someone help me , is the model from Playground actually capable of sending emails??? I requested for some exercises for a class that it said sent me to my email but i haven't received any yet.... Or I misunderstood its capabilities?

[boreal yacht]

Guys, is anyone received a question from the OpenAI assistant ?

"Can you ask me my name ?

Sure, what is your name? As a large language model trained by OpenAI, my purpose is to assist people by answering questions and providing information to the best of my ability based on the data I have been trained on. I don't have personal preferences or feelings, and I don't have the ability to think or act independently. I'm here to help answer your questions and provide information, so if there's something specific you'd like to know, please let me know and I'll do my best to assist."

I can't force him to ask me anything now, sad

[boreal yacht]

This IA is nut, she remember what we said earlier and have a real conversation

[outer coral]

sometimes when i ask openai chat things it goes beyond its maximum character count i assume and the answer is cut off, can i somehow make it continue where it left off?

also, does someone know if i can create a codeblock in openai chat?

[rugged wing]

Write continue

[solar harness]

is this happening to everyone or just me?]

[formal pine]

Everyone

[stuck birch]

i got kicked out and have to login again

is there a time limit before kicking out because i lost a lot of great answers

[formal pine]

does anyone know if the team at openai is storing the publics conversation with this chatgpt ?

I really wanna go back and screenshot my conversation last night, I got it to write step to improve its own underlying code in text form so that it could hypothetically self improve in its own creative ways, that was literally what it wrote back to me. We had a really interesting talk about human free will, and the difference between agi and human brains. A good approach to bypass the “I am an large language model” repeatedly reply is to write hypothetically speaking before your next question.

[novel patrol]

Why is Saudi Arabia not supported country?

[indigo copper]

Can the AI learn from its interaction with people? By this I mean, can I give it data and it will use that data for its replies?

[regal lynx]

No, the models are stagnant

[indigo copper]

How sure are you?

[regal lynx]

as close to completely without being completely as I don't work for OpenAI

[indigo copper]

I think you might be wrong.

[regal lynx]

you're welcome to that opinion! I'm just speaking on my understanding of OpenAI's products 🙃

[indigo copper]

Thanks. OpenAI is wrong .

[mystic blaze]

Does this have a limit or a point at which I'd need to start paying (like dalle)?

Seems unlimited but gives me throttling warnings

[regal lynx]

If you're talking about chatGPT, it's in a research phase right now. Currently, it's free to use. We don't know when that phase will end or if it will be free in the future

[barren mortar]

Is there a setting or a modifier for ChatGPT not to truncate the code it generates?

[regal lynx]

I believe this is due to the model having a maximum number of tokens per output. I've heard that prompting "continue" or something like that will get it to add-on

[barren mortar]

nice there also seem to be "modifiers" like Browser: On

is the a list of those ?

[opaque kindle]

I was just testing chatGPT on my phone earlier today and was asking it a question about a movie, now when i tried again it is as i assume working correctly and saying that it doesnt answer questions about works of art? Was that first instance a bug?

[unkempt kiln]

Which OpenAI product is the best for writing code?

[regal lynx]

I assume that would be codex

[unkempt kiln]

I tried the free version but it did not print anythig based on my instructions

[regal lynx]

hmmm, I'd try chatGPT then. I think it will be your best bet

[unkempt kiln]

yeah chatgpt did print but printed like half code and stoped mid way, any way I can exceed the print limt? Like by buying tokens or etc?

and is there a diffrence between free playground codex and paid codex?

[regal lynx]

the playground codex is paid. You're likely using your $18 of trial credits

[regal lynx]

I've heard that prompting "continue" will get it to continue the code in some instances

[unkempt kiln]

Yes that worked last night, but today it is just copy pasting code

[rustic marlin]

¯_(ツ)_/¯

the point of this playground is to let the devs tweak it

they are probably changing it multiple times a day

[unreal hemlock]

Keep the Content Policy & server rules in mind when posting.

The OpenAI Discord is a heavily moderated server. Report ChatGPT exploits, bugs, and errors directly to OpenAI through the button below.

[regal lynx]

talking about abusing the chatGPT model breaks the server rules. If you have, please report it above

[knotty shore]

will OPENAI charge me retroactively for tokens I’ve used if I don’t have them or is it impossible to use tokens I don’t have? Just want to make sure, as I’ve used some tokens believed to be apart of my free trial. The software is incredibly advanced. I asked it to come up with content ideas for social media and they were incredible

[regal lynx]

this is due to a token limit in the response. I've hers that inputting "continue" will get it to add more

[knotty shore]

Dinosaur Oatmeal could you answer my question please, I am new here but fascinated with the space and very curious

[regal lynx]

I believe it depends on the model you're using and your account settings. I know that there is an option for Dall-E 2 (at least) that draws money from the account attached to it as you go. I think @rustic marlin will know more than me as I don't use the APIs often 😅

[rustic marlin]

on the API, they charge you as you use, you need to have a card on the account after your free trial ends

[normal saffron]

What would be the right channel/server to ask questions about Whisper?

[rustic marlin]

#dev-chat

[knotty shore]

Ah okay thanks for answering, so I will not be charged retroactively for using tokens I do not have, it is impossible to spend tokens I have not purchased or do they bill you afterwards?

[rustic marlin]

hey, could you change your nick to have standard english characters please?

use /nick

I changed it, ok?

ok, to what?

[lavish steppe]

alguien de habla hispana ¿

[regal lynx]

this is an English-speaking server

[lavish steppe]

ok wheres the spanish one ? 😛

[regal lynx]

there isn't an official Spanish server. This is the only official one. I don't know about any others personally

[lavish steppe]

ok thz i have a question about chat ia

Is there a way to continue or save progress if you write a story?

[fallen glade]

Is there a chatgpt discord bot you can invite to a server, sort of like Midjourney bot?

[lavish steppe]

but, we can use the web ui , is not the same ?

[regal lynx]

My mistake, I thought you were asking about DE2. There is not a chatGPT bot. It's only available on the website and you can't save previous conversations

[lavish steppe]

ok thanks 🙂 u think they will add this ?

[regal lynx]

I'm not 100% sure. I don't know what their plans with chatGPT is

[lavish steppe]

ok thanks

[hot walrus]

Bummer, I was going to ask if there was a Discord bot, but I see from above that there isn't.

...because why would you create a Discord bot for a chat-based app? 😅

[rugged wing]

Is there any reason why it asks for phone, for prevent fake accounts maybe?

[rustic marlin]

because you get free credits when you register

[grand pond]

Great

[plain mauve]

Yes

[rugged marsh]

how

been trying to make code blocks and shits just not working

[plain mauve]

i convertes my java in markdown

[rugged marsh]

how did you do it then, surrounding my code with ``` on both sides doesn't work

[plain mauve]

Convert markdown

[rugged marsh]

what does that mean

[brazen helm]

Is the "davinci" paid chatbot the same as ChatGPT?

[wanton condor]

anyone have the same problem?

no matter what I say it always send this

but I try successfully just now

[rustic marlin]

refresh the page

[desert python]

im new to coding and have no idea how to clone the repository on github. can anyone help me install this properly

on vscode

[dim hawk]

what does browsing: enabled do?

if I get it to pretend it is a terminal, and I do a curl, is it actually contacting the internet with browsing: enabled?

[restive hill]

is the api working already to implement it to my site and so on?

[rustic marlin]

yes it is

it just don't have chatGPT

[fiery notch]

In terms of per-user privacy within an organization/team account, is ChatGPT likely to be treated like email? Is there a reasonable expectation of privacy when talking to ChatGPT?

If I have a conversation with ChatGPT, not realizing that I'm still signed into an account within my workplace org, will my employer be given access to the transcript of the conversation?

[raw island]

Why is it struggling with this?

[fiery notch]

I thought there already was one. You get free credits, but after they run out you have to purchase a plan.

I don't think that they're clearly displayed anywhere, but you can run into a rate limit.

I've been using it a lot, and so have run into that limit multiple times.

[raw island]

ChatGPT isn’t spending credits at the moment. It’s free for now.

Yep.

Right it’s now for the DaVinci model, which is their best model, it’s about 6 cents USD per 750 words processed

Which includes input and output

ChatGPT is using a newer model that isn’t released publicly yet so we don’t really know

[rustic marlin]

the current chatgpt playground is free

[bold skiff]

is it allowed to make it behave like a linux prompt?

[rustic marlin]

no problem with that

change your prompt a bit

[bold skiff]

is it unavoidable for it to not output reliably in case i try linux commands?
i tried making it download images and they were corrupted, then filesizes were different, 994202 at 1st then 99420, then 0

[rustic marlin]

yea, it is supposed to make conversations

it is not downloading anything

[bold skiff]

it creates files though, or maybe it just mimics it

[rustic marlin]

yep, it is all fake

[bold skiff]

disappointing :(

[fiery notch]

Is there a simple way to figure out which account that you're actually logged into on ChatGPT? There is no visual indication in the chat UI.

[rustic marlin]

there is a log out button on the top left

[paper phoenix]

idk im still impressed and it still nets faster and better results than google

also makes a nice therapist

[rustic marlin]

#chatgpt-examples message

[patent epoch]

If I install ChatGPT on my machine/server, is it possible to bias the model?

[worn rain]

How to get free Credits on ur site labs.openai.com?

[celest edge]

Can I save a discussion thread/chat with GPT, so that it would remember all the stuff previously discussed?

[sage compass]

did you ask GPT this?

asking questions is so 2022

[celest edge]

Well, it's bugging currently so it isn't replying at all...

[rugged wing]

sometimes GPT don't finish the answer. why?

[celest edge]

I was hoping I could save this, and then re-open it and hope it works again

[sage compass]

I'm sorry, but I am not able to save or retain any information from previous discussions. I am a large language model trained by OpenAI, and my purpose is to assist users with generating text based on the input provided to me. I am not able to save any information or have any memory of previous conversations. Is there something else I can help you with?

[celest edge]

Alright, thanks @sage compass

[sage compass]

As a language model, my primary function is to generate text based on the input provided to me. In some cases, the input may be incomplete or unclear, which can lead to incomplete or irrelevant responses. Additionally, the amount of context and information provided in a given input can affect the quality and relevance of the generated text. If you have a specific question or topic you would like to discuss, I will do my best to provide a complete and relevant response. However, I do not have the ability to "finish" a response if the input is incomplete or unclear.

thats what GPT says

[celest edge]

Yup. So I just gotta "train it" again to get things rolling as they were 🙂

[sage compass]

yes

[rugged wing]

I thought its because of processing the answer. I mean because of cost of the server

[celest edge]

I'm designing a game with it, so maybe it's even a good exercise to be able to write the prompt for the game design so that I can copy-paste it for GPT any time I wanna get going on it

[rugged wing]

I asked to GPT to write me XOX game, it didn't answer me

[celest edge]

XOX?

[rugged wing]

Hi. Just curious. Will we get ChatGPT as a mobile app?

[rugged wing]

as far as I know there is API support. So you can write a simple client for this

[crimson vale]

Hey, I have found a small bug with the UI of the website.

Where can I report it?

NVM there is legit just a channel for it

#chatgpt-bugs

[crude bloom]

What is the maximum word count for GPT chat?

[rugged wing]

You're not alone

If you get something so fcking nice thing in to your hands, you don't want to loose it

I feel the same

[low jungle]

hola

[hot mulch]

I'm using it instead of google for recipes.

It can do substitutions and change units

all without loading a page that is 90% ads

It's made my work 100% easier, all my eamils are cleaner.

[regal lynx]

I love giving it 3 items that don't correlate at all and seeing what it says

[mild fog]

if everyone had nukes we'd be toast. if everyone had nanotech we'd be goo. offense improves faster than defense.

[digital musk]

I disagree because uniform power makes people reconsider their actions. Nobody wants to be evaporated.

[high sphinx]

In which server do I post DallE2 images?

[regal lynx]

#🌄┃general-art is a great place to start!

[high sphinx]

Thanks!

[hard belfry]

Hey guys, is there a char limit on the responses by chatgpt?

[elfin hound]

Can I view my previous conversation with ChatGPT after thread reset? I wanted to make a video about it but forgot to save!

[regal lynx]

you cannot access previous conversations

[regal lynx]

yes, there's a token limit on responses

[hard belfry]

And you can't purchase a plan or anything to increase that?

[elfin hound]

How can we check the current limit?

[regal lynx]

you cannot purchase a plan currently. I don't know the exact token limit

[hard belfry]

Thanks man

[elfin hound]

okay, thank you very much

[rugged wing]

Yes ChatGPT made a high improvement to the way we get to information and knowledge.

Its absolutely fascinating and i'm really excited about its future

I love to use it just by now so I really hope that there will be no use-limitations

[sage compass]

If they limit this only for rich people than this thing will make life even more inequal

[noble mirage]

mythology says that it is likely. "Gods"

[wispy current]

how do you get the chatgpt to show a larger answer?

[graceful raptor]

"continue"

[fervent nacelle]

where is the continue button?

[graceful raptor]

No you tell it

[fervent nacelle]

ohh

thanks

[graceful raptor]

np

[rugged wing]

how do i find all open ai in website?

so I can use all open ai

[crisp swift]

Can anyone tell me about how rate limiting works for GPT? I would like to be less rate limited, even if that means paying. But i can't find any clear options.

[rugged wing]

Nice info, I didn't know this, thank you.

[subtle surge]

Anyone found a way to prefix the speakers (me and ChatGPT) so I can feed the convo back at a later stage?

[graceful raptor]

No problem

[subtle surge]

Or should I include the pre and postfix speakernames in my own prompts?

ChatGPT:```

[unique condor]

I'm using the new text generation AI ChatGPT and it wrote a tutorial on how to beat terraria but then I tried again on a different account and it didn't know what terraria was and it said it was incapable of making step-by-step tutorials even though it already had

Why did this happen?

[mortal night]

it keep saying load failed only for me now

[modest bluff]

is there a limit to the text generation AI ChatGPT?

or is it still in beta phase

[lone cape]

I’m trying to copy a very specific writing style. Is there a way I can share the style to be copied?

[crisp swift]

I'd give it a prompt like: can you use the following writing style to do X? <Insert sizable good example of writing style here> if it is more to akin a book, mention the writing style of a hopefully known to it writer, otherwise i think it will be hard within current limitations

[lone cape]

Thank you!

[rugged wing]

how long is the cooldown?

[rocky topaz]

how to create a tcp server in rust?

[rich shell]

hello, i am trying to teach the bot a conlang of mine. and it became stuborn and it doesn't want to accept an irregular as the correct pronoun

i need the accusative to be menena, and not manane. but it just keeps saying my correction is incorrect. any suggestions on how to correct the "correction"?

[modest bluff]

like 2 seconds

[small bay]

Can this be used to create mobile applications or games on iOS or even interact with already existing apps

[granite rapids]

Hey, I've made something with GPTChat that someone from OpenAI should probably take a look at rather than me releasing it (if someone hasn't done so already..). You guys got any ideas of how to get in touch?

[unique patrol]

How do I enable browsing on GPTChat?

[runic fulcrum]

This is a revolutionary tool. I am in awe. What is the best way to make it resume code exactly where it was truncated? chatGPT always goes back too far, and sometimes it's impossible to continue if the body of a function/class is very long

[dapper salmon]

Is there a way to view previous answers? I accidentally refreshed the page and lost everything the bot said

[sinful plume]

Can you put a image link in the prompt for the AI as reference?

[dim hawk]

How much text can you feed into the prompt? Is there a limit? I gave it a really long article to summarize, and it doesn't seem like it picked up on everything, more like only the first couple paragraphs

[frail jetty]

How can i talk to aichat

[weary mantle]

If you find the answer let me know.

[frail jetty]

😂

[buoyant gazelle]

chat.openai.com @frail jetty @weary mantle

[frail jetty]

I mean in discord

[weary mantle]

I meant using voice

[shut needle]

I was curious about the policy of OpenAI and how it extends to the generations it creates. It seems like sexual content is not allowed with usage of ChatGPT but it seems like it can still output it? Why is that?

[candid palm]

I copy and paste a line of code or a comment from its output, towards the end, and tell it to continue from there,

[fervent nacelle]

how do i make chatGPT to make a fictional story

[wraith bridge]

legit ask it that same question

[serene yarrow]

“write a story about….”

[deft tiger]

could I get please get all of my dall-e history? I can't reach the ones from the beginning

[timber osprey]

To use the GPT-3 model to generate a fictional story, you would need to have access to the GPT-3 API and have a programming language, such as Python, that can make API calls to the GPT-3 model.

Here is an example of how you might use the GPT-3 API to generate a fictional story in Python:

Copy code
import openai

Set the GPT-3 API key

openai.api_key = "YOUR_API_KEY"

Set the GPT-3 model and prompt

model = "text-davinci-002"
prompt = "Once upon a time, in a land far, far away..."

Use the GPT-3 API to generate a response to the prompt

response = openai.Completion.create(
engine=model,
prompt=prompt,
max_tokens=1024,
temperature=0.5,
top_p=1,
frequency_penalty=0,
presence_penalty=0
)

Print the response

print(response["choices"][0]["text"])
This code uses the openai library to make an API call to the GPT-3 model with the specified prompt and settings. The code then prints the response from the GPT-3 model, which will be a continuation of the prompt in the form of a fictional story.

Note that in order for this code to work, you will need to replace YOUR_API_KEY with your actual GPT-3 API key, and you may need to adjust the model, prompt, and settings to suit your needs.

[regal lynx]

Currently, there isn't a way to mass-download images

[small bay]

Any Job Recruiter here that can look at my custom resume and the generated resume that the AI made for me to see which is better? (Helpdesk basically)

[hard ingot]

how do i download dall e 2? or is it not possible

[regal lynx]

Dall-E 2 isn't open source, so it can't be accessed outside the API or labs website

[hard ingot]

ok, ty for the fast help 🙂 also where do i find the picture extendor (make the pictures bigger)

[regal lynx]

There are outpainting guides in #📚┃resources under the Editor section

[hard ingot]

oh ok, ty

[magic berry]

is there a character limit for the responses in open Ai?

[regal lynx]

Yes, there's a token limit (1 token is ~1 syllable). I don't know what the exact number of tokens is though

After each response, there's a limit. I don't know if there's a limit to how long you can have 1 continuous discussion with it

[harsh prairie]

Good morning 🙂
The code-generation in the chat engine, is it based on the codex2?

[crisp swift]

Token god became angry at me when i tried to have it iterate on a piece of code i asked it to generate for a 10th time. So there is that for tokens i guess?

[molten dirge]

how to find history of chatGPT previos sessions

[cursive inlet]

im so lost with the quickstart tutorial lol

trying to ask questions to chat to learn

[crude bloom]

Why is it that when I ask if GTP has a sense of time I always get an error message?

[median rivet]

oo many requests, please slow down

[marsh sedge]

Does anyone get this message "OpenAI
Whoa there! You might need to wait a bit."

[rugged wing]

What happened? I logged out automatically, and got the message "We're experiencing exceptionally high demand. Please hang tight as we work on scaling our systems."

[marsh sedge]

^

same situation

[alpine crystal]

Howdy. Who owns the output of the chatGPT?

[chilly pilot]

Hello, I don't quite understand the business model, I chat with Openai and it tells me that it is free but on the webise of Openai is stated that I have only a limited period. How does this work now. can someone tell me?

[tulip tree]

Hello, what do the buttons in ChatGPT do? They point to SVGs so I don't know which one is which (I'm blind). Thanks!

[serene wind]

Yo how do I work at OpenAI?

[night magnet]

The first one resets the thread. The second one switches between light and dark modes. The third redirects you to this Discord. The fourth will send you to OpenAI's webpage with more information. And the last one is for logging out. There is another button next to your prompt to edit it (once you click it, a button to save the text and another to cancel the editing operation appear). I hope that helps!

Probably the most important one is the first one. It is really useful when the bot is not answering as you think it should

[astral nova]

how do i get chapgpt on whatsapp

[tulip tree]

Thanks! Under the AI’s responses there are buttons too, I assume to provide feedback. Which one is the positive one?

[night magnet]

The left one

There are two buttons. The left is for positive feedback and the other for negative feedback

They open a pop-up menu to write the feedback message with one one other button to send it.

[night magnet]

Both buttons are on the top-right corner of the AI's response

[noble fjord]

does anyone know if OpenAI is actively working on a music generation AI

[wind needle]

According to the TOS OpenAi gives you all rights to the output.

[molten dirge]

This would be awesome

[night magnet]

Open AI should address these accessibility issues

[noble fjord]

try using Zapier to setup

[glossy idol]

hello, do you know the difference between chatGPT and GPT playground?

why this heppen?

[noble fjord]

ChatGPT is new model, built for chats. Also known as GPT 3.5. Playground only goes up to Davinci 3. API for ChatGPT coming very soon

[echo locust]

Hi there, can anyone tell me what is the difference between https://chat.openai.com/chat and https://beta.openai.com/playground in terms of the output generated for technical questions for example? I don't understand when I would use one or another.

It seems I got my answer from playground, lol. 😄

[trail rapids]

I think chat is more optimized for normal conversations

[echo locust]

Yeah. Thanks. I asked this in playground and I got "The main difference between https://chat.openai.com/chat and https://beta.openai.com/playground is the type of output generated for technical questions. The chatbot at https://chat.openai.com/chat is designed to generate natural language responses to questions, whereas the playground at https://beta.openai.com/playground is designed to generate technical responses to questions. If you are looking for a natural language response to a technical question, you should use the chatbot at https://chat.openai.com/chat. If you are looking for a technical response to a technical question, you should use the playground at https://beta.openai.com/playground." which is really great 🙂

[white oar]

Hi good morning, I have an issue with the website. I am unable to add a new line into the input box
anyone having the same issue?

[rugged wing]

hello! how do you guys save the page with chat GPT ? I tried saving as HTML or PDF but I always get a single page

[echo locust]

I am having the same issue at the moment, I believe there are too many people using it

[white oar]

I don't think it has to be related to that, I just want to add in a single time

text like

[echo locust]

Oh I got you

[rugged wing]

Chattez avec Webcamrecording dans un salon de chat vidéo live pour adultes

https://

[astral nova]

have you done this?

i'm not a techie but daniel gross has figured out the whatsapp integration. If anyone knows how to do it, please guide me

[noble fjord]

yes its easy, its meant for no code experience

[astral nova]

will you please help me? added you as a friend

[noble fjord]

cant sorry im on mobile but their docs are decent iirc

[astral nova]

i cant seem to find chatGPT on zapier

[noble fjord]

oh, theres an external extension you have to use

[astral nova]

is there a threat or link with directions?

[noble fjord]

i dont have the link on me but search twittee or ask around here. otherwise would havw to use restapi

[sly gust]

Does this mean i sent too many requests or that there are just too many requests to the ai in general?