yantao0527
yantao0527
I recommend a [youtube video](https://youtu.be/Ix9WIZpArm0) to learn how to use OpenAI's new GPT-4 api to 'chat' with and analyze multiple PDF files.
@merrymercy I understand adding a new field into Conversation class that is already complicated is a bad idea. Let me think about how to reduce the complexity of Conversation class....
In order to reduce the complexity, I split Conversion class into two parts, Prompt class and Session class, in fastchat/session.py.
Message class represents the life cycle of a interactive that occurs from user input to system output, also includes moderation, http request bot, vote result, or other performance data. So...
Sure. I just provided my thought to reference after I learned project code. I'm glad to discuss it with you and look forward your refactor.
Which strategy should we use? - Sanitize user input [PR#571](https://github.com/lm-sys/FastChat/pull/571) - Escape your output instead
When I input the prompt "<img onerror=alert(1) src=#></img>" it should show the same content in the chatbox. But the attribute "onerror=alert()" is sanitized and a broken image is showed in...