Greetings,
I use the ChatGPT interface quite frequently. Sometimes, I create amazing prompts. Some of them turn out to be exceptionally perfect, they are generally made up of feedback prompts in the scope of meta-prompts. Later on, I switch to another topic, and these prompts just get lost 
Do you experience this too?
You are right, we can save them with some auxiliary documentation applications or create a system template for our prompts; this helps to some extent. We should keep the entire chat for the meta-prompts.
If the chat groups on the left side that hold the conversation flow can be user-defined and indexed, and even bookmarked, we could prevent this chaos on the left side.
Additionally, I’ve become quite accustomed to GPT 3.5 being green, GPT 4 being purple. These colors could be used to differentiate models by coloring the icons of the chat groups.
What do you think?
Could the GPT team include such an improvement in your business plan?
Thanks,
1 Like
Welcome to the Forum!
To your first point, the problem is one of context length, there is a fixed amount of text that can be included with each call to the AI, so that limits how much of the past chat text can be included, the new GPT-3.5-Turbo-16k API allows 4 times the chat context length as the prior ChatGPT, but this requires that you use the API and the Playground OpenAI Platform rather than ChatGPT.
Not sure what you mean by coloured icons, I thought they were already coloured.
1 Like
Hello,
Actually, I am aware of the context length, I may have conveyed it incorrectly. I want to try again with a simpler expression; there should be a directory system where the user can create their own groups (parents) using chat groups (listed on the left). This reduces complexity on the left, reduces duplicates. The user should categorize the chats they have with GPT themselves.
To give an example:
I have a few chat groups about Docker, one of them is about Dockerfile writing, another one contains information about Docker Compose. I processed embedded information within these chat groups that would reveal GPT’s capabilities and I can get the outputs exactly as I want. I don’t want to lose these chat groups.
Let’s think like this; we worked a bit, we got help from GPT, we used these Docker-related groups. Then time passed, we talked about other things. We’ve talked so much that I can’t reach the chat groups on the left where I provided embedded information about Docker that can give me the outputs I want. I wonder under which title were these chat conversations related to Docker? My time is limited, should I search which one it is among these groups? Or should I create a new chat group?
Mostly, I start a new conversation/chat. This situation includes taking notes of the input templates and the chat.
If it is still not understood; you can think about the nested categorization event in the bookmark system in browsers.
This is a situation I am experiencing; I have no idea how other users, and you use GPT. Would it be a suitable approach for your own situation? Because I use it quite intensively, it would be nice to have such a thing. Still, we can do a survey in the community, action can be taken according to everyone’s opinion.
Thanks,
Best wishes.
I really don’t understand what exactly it is you’re wanting.
Why don’t you just use the bookmark system in Chrome?