Add FLAS to leaderboard#152
Open
Lunamos wants to merge 1 commit into
Open
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Add a "Beyond rank-1" subsection under the rank-1 steering leaderboard in the README, with FLAS (Flow-based Activation Steering) as a first entry
Why a separate section: FLAS is not a rank-1 intervention. Rather than computing one steering vector per prompt and adding it at every position (as HyperSteer does), it integrates a learned velocity field over multiple steps, where the displacement is re-evaluated from the current activation at each step and differs across token positions. Recent methods like AcT and K-Steering also relax these restrictions, so this class may keep growing.
Scores follow the main table's setting (held-in harmonic mean, layer 20, Concept16k): 1.025 on Gemma-2-2B-IT, 1.185 on Gemma-2-9B-IT, at a single fixed flow time with no per-concept tuning
Left
index.htmluntouched for now — happy to mirror the section there too if you'd likeWebsite: https://flas-ai.github.io
Paper: https://arxiv.org/abs/2605.05892
Code: https://github.com/flas-ai/FLAS
Models: https://huggingface.co/flas-ai
Possible follow-ups (if there's interest)