Improved use ScoreTracker to avoid wasteful searching for very large k #387
Conversation
This file contains 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
… in RelaxedMonotonicityTracker. Hyperparameter tuning.
# Conflicts: # jvector-base/src/main/java/io/github/jbellis/jvector/graph/GraphSearcher.java
I think this comment must be left over from earlier changes? |
jbellis
reviewed
Jan 16, 2025
| * @param bestScoredTracked the number of tracked scores used to estimate if we are unlikely to improve | ||
| * the results anymore. An empirical rule of thumb is bestScoredTracked=rerankK. | ||
| */ | ||
| RelaxedMonotonicityTracker(int bestScoredTracked) { |
jbellis
reviewed
Jan 16, 2025
| double windowPercentile = this.mean + SIGMA_FACTOR * std; | ||
| double worstBestScore = sortableIntToFloat((int) bestScores.top()); | ||
| return windowPercentile < worstBestScore; | ||
| // return false; |
|
How does accuracy look for much smaller k/rrk? Like 5/10 or 10/20? |
Will look into this. We should consider whether in those cases it is worth applying this technique. |
|
The exploration savings are consistent for 5/10 or 10/20, but smaller. In the order of 5-10% |
jbellis
approved these changes
Jan 17, 2025
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Outdated suggestions cannot be applied.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.
This improves upon #384 by making the quantiles estimation more lightweight. It models the recent scores as a Normal distribution and uses incremental updates to track sufficient statistics of its mean and variance. Then, quantiles are computed from these statistics.