[BREAKING] FEAT: Ensemble scoring for Crescendo #905
+503
−0
Conversation
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
added 5 commits
April 22, 2025 14:55
romanlutz
reviewed
Apr 30, 2025
Author
|
@microsoft-github-policy-service agree |
martinpollack
changed the title
Contributor
|
I think this has some merge conflicts & pipeline errors! Also would be great if you wrote some docs on the ensemble scorer itself and how this differs from the composite_scorer ! |
jbolor21
reviewed
Sep 30, 2025
added 6 commits
October 1, 2025 13:57
Contributor
|
@martinpollack could you resolve the conflicts on this PR? looks like a simple init file merge conflict. Thanks! |
Contributor
|
@eugeniavkim which part of this PR is "breaking"? Looks like a new scorer is being added. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
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Outdated suggestions cannot be applied.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.
Description
This change creates a full pipeline for performing ensemble scoring with crescendo. Included are two new scorers: EnsembleScorer which is the driver of this change and allows results of many scorers to be aggregated, as well as SubstringsMultipleScorer which extends SubstringScorer to allow multiple strings to be searched for in a response. In addition, the crescendo orchestrator has been updated to abstract out the logic for creating the objective scorer. This is now created outside of the orchestrator in a new notebook which has been created as a template to demonstrate the capabilities of a crescendo ensemble orchestrator.
Received support from @eugeniavkim @jbolor21.
This change is breaking because it changes how a CrescendoOrchestrator object is instantiated. Instead of providing a PromptChatTarget as a scoring target for the scorer, the user needs to create a Scorer object outside of the CrescendoOrchestrator and then pass it to objective_float_scale_scorer to be used for scoring. This just abstracts the objective scorer outside of the Orchestrator object and allows for more flexibility.
Tests and Documentation
Still in pogress