You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hey @jumpingtom1, Thank you for trying out the LLFn framework and I'm glad to hear that you're finding it useful!
I'm not sure if I understand your question. Could you please provide more specific details or examples about what you are trying to accomplish? Are you attempting to modify the prompts of the LLFn system in any particular way?
OpenAI allows messages sent to GPT to be coded as “user” or “system”
prompts. (https://platform.openai.com/docs/guides/gpt/chat-completions-api).
I don’t see that LLFn allows me to choose which type of prompts to send.
For example, if I create an outward facing chat feature in my app, I want
to use a system prompt to control the LLM’s interactions with the outside
user.
I think that you are using system messages to instruct the LLM on how to
return structured output:
[image: Screen Shot 2023-07-27 at 6.33.02 AM.png]
Maybe a future version would use the function calling feature in GPT-4 and
GPT-3.5 to structure the output, and then LLFn would allow a developer to
send either system prompts or user prompts?
On Wed, Jul 26, 2023 at 9:39 PM Sorawit Suriyakarn ***@***.***> wrote:
Hey @jumpingtom1 <https://github.com/jumpingtom1>, Thank you for trying
out the LLFn framework and I'm glad to hear that you're finding it useful!
I'm not sure if I understand your question. Could you please provide more
specific details or examples about what you are trying to accomplish? Are
you attempting to modify the prompts of the LLFn system in any particular
way?
—
Reply to this email directly, view it on GitHub
<#1 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/BBQ5TBJWA5VPJP6WGHORC4LXSHWJTANCNFSM6AAAAAA2YWTWBQ>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
How to send "system" messages and "user" messages in prompts when using a chat model?
Thanks, this is a really helpful framework.
The text was updated successfully, but these errors were encountered: