Activation steering produces synthetic safety-violating data that improves downstream classifiers over prompting on most tested concepts when a harmonic mean of alignment, coherence, and diversity is optimized.
Grains: Gradient-based attribution for inference-time steering of llms and vlms.CoRR, abs/2507.18043, 2025a
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
citation-role summary
dataset 1
citation-polarity summary
years
2026 4verdicts
UNVERDICTED 4roles
dataset 1polarities
use dataset 1representative citing papers
Continuous Interpretive Steering and the GraSD dataset reveal that LLMs encode graded sensitivity to scalar diversity in their internal representations, recoverable via controlled activation interventions.
The Master Key Hypothesis states that capabilities are low-dimensional directions transferable across models through linear subspace alignment, with UNLOCK demonstrating gains such as 12.1% accuracy improvement on MATH when transferring CoT from 14B to 7B models.
The survey organizes mechanistic interpretability techniques into a Locate-Steer-Improve framework to enable actionable improvements in LLM alignment, capability, and efficiency.
citing papers explorer
-
Activation Steering for Synthetic Data Generation: The Role of Diversity in Downstream Safety Detection
Activation steering produces synthetic safety-violating data that improves downstream classifiers over prompting on most tested concepts when a harmonic mean of alignment, coherence, and diversity is optimized.
-
Continuous Interpretive Steering for Scalar Diversity
Continuous Interpretive Steering and the GraSD dataset reveal that LLMs encode graded sensitivity to scalar diversity in their internal representations, recoverable via controlled activation interventions.
-
The Master Key Hypothesis: Unlocking Cross-Model Capability Transfer via Linear Subspace Alignment
The Master Key Hypothesis states that capabilities are low-dimensional directions transferable across models through linear subspace alignment, with UNLOCK demonstrating gains such as 12.1% accuracy improvement on MATH when transferring CoT from 14B to 7B models.
-
Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models
The survey organizes mechanistic interpretability techniques into a Locate-Steer-Improve framework to enable actionable improvements in LLM alignment, capability, and efficiency.