Calibrating Lightweight Sparse Autoencoder Feature Steering

Abstract:Sparse autoencoders (SAEs) can enable inference-time topic steering by modifying latent feature activations, but existing steering methods often fail when target-aligned features are not identified or are modified at the wrong scale. We introduce \textsc{ContrastiveSteer} to address these respective failure modes. First, features are scored by how much more strongly they activate on target-domain text than on general text. Second, steering strength is set using a model-specific calibration. We also introduce \emph{contamination}, a complementary heuristic that measures whether post-steering activations emphasize weakly aligned features, thus indicating off-target or nonsensical outputs. In our evaluations, across multiple LLM families and SAEs, \textsc{ContrastiveSteer} improves topic alignment across domains up to 3.9$\times$ over unsteered and steering baselines, and raises target-domain classification by up to 93\%. Our activation-scale normalization method also recovers at least 93\% of optimal domain-score gains across all tested model families, and contamination distinguishes successful steering, under-steering, and over-steering collapse.

Computation and Language (cs.CL); Artificial Intelligence (cs.AI)

arXiv:2506.12576 [cs.CL]

(or arXiv:2506.12576v3 [cs.CL] for this version)

https://doi.org/10.48550/arXiv.2506.12576 Focus to learn more arXiv-issued DOI via DataCite

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.