[Feature] Support model-bound prediction postprocessor, use it in Claude #268
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
Leymore
approved these changes
Aug 25, 2023
go-with-me000
pushed a commit
to go-with-me000/opencompass
that referenced
this pull request
Oct 9, 2023
…compass#268) * [Feature] Support model-bound text postprocessor, add claude as an example * update * update * minor fix --------- Co-authored-by: zhoufengzhe <zhoufengzhe@pjlab.org.cn>
liuyaox
pushed a commit
to liuyaox/opencompass
that referenced
this pull request
Jun 26, 2024
…ude (open-compass#268) * [Feature] Support model-bound text postprocessor, add claude as an example * update * update * minor fix --------- Co-authored-by: zhoufengzhe <zhoufengzhe@pjlab.org.cn>
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Outdated suggestions cannot be applied.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
Different models have different output styles, sometimes it's hard to share a universal postprocessor across all the models, which may underestimate some outlier's performance.
Modification
This PR adds a
pred_postprocessorparameter tomodel's config, where users can customize postprocessors with respect to each datasets via a dataset_abbreviation -> postprocessor construction dict. The configuration from model side has the higher priority over the dataset-side.BC-breaking (Optional)
None