I encountered a problem. I want to let gpt convert it into an interface document according to the curl situation, and give it to me in the form of original markdown text, but it always seems to render errors. Can anyone fix it?
I don’t know of a way to do it within ChatGPT, but you can go to platform.openai.com and use the Playground API to run the same query, the API is not connected to ChatGPT so you will need to add a payment method to make use of it, but that Playground system will output raw text that will include the markdown.
I am using gpt3.5, it gives me markdown texts, but it renders parts of them. So the answer seems weird, some is plaint markdown, some is rendered markdown. I am not able to do the payment, would this bug will be fixed in near future?
@Foxalabs can you please help me out from the trap of markdown? I have to ask gpt forget my prompts before I really have a new prompt. If I don’t ask to forget, GPT will repeatly reply me with markdown
What you can do if complex markdown is the end result is to tell the AI:
- markdown rendering has been disabled in your user’s display, user will only see the exact text that you send. Therefore, produce no markdown for the sake of presentation;
- the prohibition on producing markdown extends to code blocks. No backticks (```) are allowed, even when producing computer code;
- The user may request a markdown document they can use elsewhere. In that case, the entirety of your reply will be only a document using markdown, and what you create will be for reuse by a strict markdown-to-typesetting library;
give me an example about amqp
This instantly transitions the output from being a display of communications between sender and recipient to being an explanation.
The response style instructions can be moved into “customize ChatGPT”, which you can find in the settings menu when clicking your user name.