-
Notifications
You must be signed in to change notification settings - Fork 12
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
Response engineering #8
Comments
One thought is to take masterOutput from localStorage and do something with it in a way that pulls key phrase or words to include it in the input prompt. That way we can compress the stream to only what is needed. maybe what ever the javascript equivalent to grep -v noise_words. Basically anything not a noun, is what I'm thinking.. |
Ill probably wait for additional api release as this matures to figure this out with best path forward. |
At the moment, Prompts are engineered with the initial prompt to always be sent, along with the last response returned and the next passed input. Works pretty well with simple follow ups:
-- Who was was Zeus?
-- Did he have kids?
-- Did he have wives?
and can get maybe 5 or so questions in before the AI seems to get lost.
Pondering: Should I attempt to send the entire conversation from the start?
Possible issue: Token length will get exponentially smaller for new inputs/responses since token takes into both input and output.
The text was updated successfully, but these errors were encountered: