Update CRIMSON metric to align with up-to-date CRIMSON implementation#21
Open
MohammedSB wants to merge 1 commit intojbdel:mainfrom
Open
Update CRIMSON metric to align with up-to-date CRIMSON implementation#21MohammedSB wants to merge 1 commit intojbdel:mainfrom
MohammedSB wants to merge 1 commit intojbdel:mainfrom
Conversation
- Match HF model init: padding_side="left", generation_config detection, flash attention 2 support, repetition_penalty=1.1 - Auto-disable guidelines for default MedGemma HF model (trained without them) - Port robust JSON parsing from standalone (handles orphan keys, bad escapes, unescaped quotes, smart quotes, malformed arrays) - Fix scoring fallback defaults to match standalone (0.25 for unknown significance instead of 0.0) - Update default HF model to rajpurkarlab/medgemma-4b-it-crimson - Add cache_dir parameter for controlling HF model cache location - Add CRIMSON to README acknowledgments I also saw that you added the repetition_penalty=1.1, which is good for helping the model not get stuck in the self-repetitive loop, which it sometimes does. This is especially true if the inputs are not formatting as expected. Do you have any tips on how to improve fine-tuning to avoid this?
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
I also saw that you added the repetition_penalty=1.1, which is good for helping the model not get stuck in the self-repetitive loop, which it sometimes does. This is especially true if the inputs are not formatting as expected. Do you have any tips on how to improve fine-tuning to avoid this?