Add distance threshold optimizer classes #292
Conversation
rbs333
commented
Feb 27, 2025
rbs333
commented
- add base, router, and cache threshold optimizer
- uses ranx to support eval metrics
- class is designed to receive a optimizer class in the case where someone wants to supply their own without us having to make a release to support something custom
rbs333
requested review from
bsbodden,
abrookins,
tylerhutcherson and
justin-cechmanek
tylerhutcherson
changed the title
| PRECISION = "precision" | ||
| RECALL = "recall" | ||
|
|
||
| def __str__(self) -> str: |
There was a problem hiding this comment.
I think EvalMetric(str, Enum) would do the trick? See how we handle other enums in the lib
| return max(metrics.items(), key=lambda x: (x[1]["score"], -x[0]))[0] | ||
|
|
||
|
|
||
| def _grid_search_opt_cache( |
There was a problem hiding this comment.
any particular reason these aren't members of the class?
There was a problem hiding this comment.
My thought process is this shows that you can pass in any function that meets this definition not dependant on any class state if you wanted to go a more custom route. It's a little atypical though since it is still an internal method that doesn't really make sense to expose outside of this context.
There was a problem hiding this comment.
I would add imports here similar to vectorizers so we can do from redisvl.utils.threshold_optimizer import CacheThresholdOptimizer
Might also consider just naming it redisvl.utils.optimize to follow the same grammar as redisvl.utils.vectorize
