FT.SEARCH quantised embeddings support #4094
samueldashadrach
started this conversation in
Ideas
Replies: 1 comment
-
|
@romange tagging for visibility |
Beta Was this translation helpful? Give feedback.
0 replies
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.
-
Self-explanatory.
1-bit embeddings perform nearly as good as 4-byte embeddings while saving lots of RAM. Please consider supporting all levels of quantisation (1-bit 2-bit 4-bit 1-byte 2-byte 4-byte). If you can't support all, my personal recommendation is support the lower end (1-bit, 2-bit).
Beta Was this translation helpful? Give feedback.
All reactions