Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Functionary small 2.4 is ignoring the system prompt. #164

Open
gunpal5 opened this issue Apr 22, 2024 · 3 comments
Open

Functionary small 2.4 is ignoring the system prompt. #164

gunpal5 opened this issue Apr 22, 2024 · 3 comments

Comments

@gunpal5
Copy link
Contributor

gunpal5 commented Apr 22, 2024

Hello,

I was testing the Functionary Small 2.4 model for its capabilities. I found a glitch or bug in the model. I tried to alter the behavior of the model with system prompt. but model is ignoring the instruction in system prompt.

I am using OpenAI client btw.

Thanks,
Gunpal Jain

@gunpal5
Copy link
Contributor Author

gunpal5 commented Apr 27, 2024

Any comments??

@jeffreymeetkai
Copy link
Collaborator

Hi, unfortunately, we have not trained our models with a variety of system prompts. The focus for this model (v2.4) is function-calling and code-generation. Thus, it may not follow instructions in your own system prompts that well.

@gunpal5
Copy link
Contributor Author

gunpal5 commented Apr 29, 2024

What I wanted to do is use these models for AI Agents reasoning engine. I want to build a smart AI agent using functionary. which will use only to optimize the prompt for the second/Main AI model to write content.

What it needs to do is analyze the prompts and decide whether to call a function or not.

  1. if there's no function calling involves it needs to write the original prompt back.
  2. if there's a function calling involves, it needs to rewrite the original prompt with the function response. e.g.
<context>
function response
</context>

original prompt

I need to learn how to fine tune it to do the same. I think small model should be fine for this type of scenario.

Edit:
I'll try to hardcode the behavior I want in the block diagram

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants