[Feature] Added Bradley-Terry subjective evaluation method to wildbench, alpacaeval, and compassarena datasets #1791
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
…d bradleyterry subjective evaluation method for wildbench, alpacaeval, and compassarena datasets; added all_scores output files for reference in CompassArenaBradleyTerrySummarizer;
MaiziXiao
approved these changes
Dec 31, 2024
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Outdated suggestions cannot be applied.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.
Motivation
Added Bradley-Terry subjective evaluation method to wildbench, alpacaeval, and compassarena datasets.
Modification
base_models_abbrs(from dataset config) to references during subjective evaluation (passed to references inLMEvaluator). This can be used to determine the base models in the post-processor function for pairwise subjective comparisons.base_modelsin regular (non-Bradley-Terry) subjective evaluation post-processors for the following datasets: wildbench, alpacaeval, compassarena. Previously, the base model for comparison is assumed to beanswer1in the first record of references (references[0]["answer1"]), but this might not necessarily hold ifinfer_orderis set to "random". Using thebase_modelspassed from the dataset configuration guarantees that the correct base models are used.all_scoressummary files for reference inCompassArenaBradleyTerrySummarizer.BC-breaking (Optional)
No breaking changes.
Use cases (Optional)
Perform subjective evaluation using the Bradley-Terry method with the following command:
More details about the Bradley-Terry evaluation method in: opencompass/configs/datasets/subjective/compass_arena_subjective_bench/README_pairwise_bt.md