docs: add UDS (Unlearning Depth Score) to evaluation metrics links#197
Open
gnueaj wants to merge 1 commit into
Open
docs: add UDS (Unlearning Depth Score) to evaluation metrics links#197gnueaj wants to merge 1 commit into
gnueaj wants to merge 1 commit into
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Adds a link to Unlearning Depth Score (UDS), a white-box metric that measures
unlearning depth via two-stage activation patching on internal hidden states,
evaluated on TOFU forget10 using open-unlearning's Llama-3.2-1B-Instruct
checkpoints (150 unlearned models, 8 methods).
Project Page: https://gnueaj.github.io/unlearning-depth-score/
Paper: https://arxiv.org/abs/2605.24614
Code: https://github.com/gnueaj/unlearning-depth-score
It's the first mechanistic metric for unlearning depth, and we thought it'd be
a nice addition to this list. Thank you for maintaining such a great resource
for the unlearning community 😊
Best,
Jaeung