Evaluating the Semantic Specificity of Representation Steering in Language Models
Mirrored from arXiv — NLP / Computation & Language for archival readability. Support the source by reading on the original site.
Computer Science > Computation and Language
Title:Evaluating the Semantic Specificity of Representation Steering in Language Models
Abstract:Localized Representation Steering (LRS) is widely used to correct reasoning pathologies in large language models. However, standard benchmark evaluations can easily be fooled by superficial label overrides, creating a false impression of reasoning circuit repairs. In this work, we propose Cross-Rule Transfer (CRT), a diagnostic framework that audits representational interventions by evaluating them on rule families where the model is natively competent. Evaluating late-layer LRS for a widespread logical failure, contradiction blindness, reveals that the intervention merely injects a global label bias: applying the steering vector to rules the model already handles correctly (99.6% baseline) degrades performance to 40.4% by forcing false contradiction predictions. We support this diagnosis with four complementary controls (direct logit bias equivalence, control vector label-flipping, cross-model grafting, and early-layer steering checks), providing a rigorous methodology to distinguish genuine reasoning repairs from superficial label overrides.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2608.29431 [cs.CL] |
| (or arXiv:2608.29431v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2608.29431
arXiv-issued DOI via DataCite (pending registration)
|
Access Paper:
- View PDF
- HTML (experimental)
- TeX Source
References & Citations
Bibliographic and Citation Tools
Code, Data and Media Associated with this Article
Demos
Recommenders and Search Tools
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
More from arXiv — NLP / Computation & Language
-
Learn Your Own Thoughts: Abstract Token Curriculum
Sep 18
-
Uni-LaDiR: Latent Diffusion Unifies Multimodal Reasoning
Sep 18
-
MATCH: Model-Aware Tool Learning with Curriculum Scheduling and Hierarchically Gated Rewards
Sep 18
-
Modality Discrepancy Transformer for Ambivalence and Hesitancy Recognition
Sep 18
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.