Is this a joke?

[oak idol]

So. I started to pay for the model to have a better experience but the interactions are getting slower and slower, the model even freezes on the web and to make matters worse, now it doesn't allow me to send any more messages in the chat that I had already trained with the task to be performed.
"You have reached the maximum duration for this conversation, but you can continue talking by starting a new chat"
I mean... Is not cheap. Specially for latinamerican users. So at least you guys can have some REAL online support instead of a complains forum.

[wise sandal]

@oak idol

Please follow the posting guidelines.

For better assistance from the community or an OpenAI team member, please follow the posting guidelines by clicking the button below.

➜ Need to contact OpenAI support? Visit https://help.openai.com

[cinder tapir]

It's surely not intended as a joke, but some ways some of us use the model might not be expected or optimized for yet.

I'm just a community member like yourself.

I do enjoy prompt engineering; figuring out how to get what you want from the model.

I've explored teaching it inside conversations.

Conversations do have a limit, they can only be so long, but that can be pretty long.

It might be worth trying to figure out how to 'more quickly and briefly' teach the model exactly what you want it to do.

That can be explored in a few ways, including asking the model to help you understand what it understands about what you want - the model might be able to help you guide it (in a new chat) to do what you want in just 1-5 prompts, then the rest of the conversation can be it doing the work you want done.

I only understand English, which might make it complex to help you if you need to use only Spanish, but we can use #prompt-engineering to discuss in English what you want the model to do and how to explain this to the model, if you'd like.

[oak idol]

Hi mate! Thank you for your answer. I acually trained the model with 2 promtps and a document last night. Is not a long chat. But I think I might have some problems on my account. For example the issue with the canvas model. I have to specify to ChatGPT to open a canvas every single time (btw... thanks for that tip). Is not a big problem anyway. But I was really pissed when that message came out of nowhere. I have other conversations with the same model and are longer (a lot) and never had a problem.

[wise sandal]

Is this issue solved?

If you believe this issue is now ready to be closed, please press the button below.

[oak idol]

Oh. And a little detail. That chat works fine on my phone. But it doesn't work in the website. And other weird thing is it doesn't update with the content I sent from my phone.

[cinder tapir]

Welcome. I'm not sure if Canvas may have a smaller 'conversation length' than non-Canvas chats - it might if the model can access every previous version of Canvas, or if the Canvas is part of the conversation being checked every input (could be a lot of extra tokens). I haven't used Canvas much yet, or tried to have any really long chats with it.

Excellent that you got 2 prompts and a document.

If the document fits inside the prompt, you could probably try an initial prompt of:

[Open Canvas and place the following text into it:

[text]

Make these changes to that text

[whatever you want done]]

But yeah, maybe with every input the entire canvas (and maybe the version history too) is being fed as part of the input, that might really eat tokens and cause a very short conversation limit maybe - I'm guessing.