How to limit AI response to max 300 characters

Hi bhavingohil, are you referring to the chat bot? You can reduce your max token maybe?

Thank you, Senol.

Yes, I’m using a chat bot widget, and I’ve set the maximum token to 65. So I’m anticipating a maximum of 300 characters of AI response. However, there is no consistency in the limit, and it has exceeding 500 to 500 characters.

How about reducing the maximum tokens to around 30? Would that help?

Hi Senol,

Annoying, before I update to token limit to 30 to limit the AI response; new issue appearing from the AI response that says: “You have reached your token limit.” Can you please advice why this happened?
And can you please explain what’s the difference between the account type in below 2 account interface:
1. https://chat.openai.com/
2. https://platform.openai.com/account/usage

Hi @bhavingohil

If you are receiving “You have reached your token limit.” it mean you setup a token limit. Please check Token Handling tab and adjust that limit.

1. That is chatgpt
2. That one is for calling API.

Thank you Senol; changing the Token Handling settings resolved my token limit issue.

From my original concern, limiting token cutting down the AI response in-between, which is inappropriate. So, how to set the limit of AI response? let’s say I want AI to answer in max 300 characters.

Hi Senol, can you please respond?

Hi Senol,

Can you please check my query and update?

Hi @bhavingohil is lowering max token not solving this?

Nope.

Can you please respond?

Hello @bhavingohil,

You may want to use the “additional context” field under the context tab to guide your bot towards responding with a maximum of 300 characters. You could instruct it with something like, “Please ensure your responses are succinct and informative. The length of responses should not exceed 300 characters, this includes spaces and punctuation.”

Please note that our plugin’s role is primarily to send prompts to OpenAI and receive responses. Control over the response is beyond the plugin’s capabilities, as the responses are produced directly by OpenAI API services. This is an aspect our plugin doesn’t manage.

Moreover, if the responses are not as per your expectations, continue to refine and improve your prompts. Through trial and error, you will learn more about how the API reacts to different inputs, enabling you to yield the desired outputs. Here some best practices: https://platform.openai.com/docs/guides/gpt-best-practices

Thanks Senol, I’ll try.