Sizing Context Data (Stuffing the Prompt) LastMaxTokenSizeContentPurger #2568
jasonparallel
started this conversation in
General
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
What is the plan for sizing context data to meet the token window?
I see LastMaxTokenSizeContentPurger was added in M1 but it does not appear to be used (or mentioned in the docs).
Any plans to shorten documents vs excluding documents? The functionality I wrote before noticing spring-ai utilizes a sentence detector to shorten documents by whole sentences.
Beta Was this translation helpful? Give feedback.
All reactions