Repository navigation
Build indexes using compressed vectors - #244
Merged
Merged
Conversation
added 9 commits
March 8, 2024 03:42
…n and Reranker to top-level
jkni
self-requested a review
March 8, 2024 17:37
added 19 commits
March 8, 2024 16:29
… vectors to (from) disk as POC
jbellis
force-pushed
the
live-compression-2
branch
from
March 22, 2024 14:57
eb19107 to
ff29179
Compare
jbellis
marked this pull request as ready for review
March 25, 2024 15:55
jkni
reviewed
Mar 25, 2024
jkni
left a comment
Contributor
There was a problem hiding this comment.
Few questions/comments -- I didn't look at Bench changes because of your comment suggesting they wouldn't be merged as-is. Is that still true?
added 9 commits
March 26, 2024 09:23
…eletedNeighbors takes care of the rest
…ch of CNS objects using the old one, and even if you updated all of those, you'd have to do it while no changes were in progress
…ike "actually everything is deleted"
jbellis
force-pushed
the
live-compression-2
branch
from
March 26, 2024 17:33
1bc5f53 to
5db274c
Compare
jbellis
force-pushed
the
live-compression-2
branch
from
March 26, 2024 18:15
5db274c to
115ef56
Compare
Contributor
|
LGTM! I took the liberty of adding missing licenses for CI, so if you force-push, make sure you pick up those. |
jkni
approved these changes
Mar 27, 2024
Contributor
Author
|
Oops, I thought I did pull, sorry. |
added 3 commits
March 27, 2024 08:49
Open
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Changes to Bench are quick-and-dirty and not intended to be committed as is.
BuildScoreProvider encapsulates build-time similarity APIs. The main changes to the actual logic are in ConcurrentNeighborSet to perform two-pass diversity checks to reduce the full-precision comparisons required.
Note that
This is important for performance (2x neighbors = 2x slower searches) and it means we don't need to do crazy juggling around "now that i'm writing to disk i'm going to recompute all the scores."