AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Local Linearity of LLMs Enables Activation Steering via Model-Based Linear Optimal Control

Julian Skifstad, Xinyue Annie Yang, Glen Chou

ICML 2026 · Architectures and efficiency › RNNs and normalization · approach: Layer-wise analysis of language models · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use A-LQR with linear feature setpoint instead of baseline steering methods for toxicityA-LQR · RealToxicityPromptstoxic continuations (%): 5.14 -> 0.12 (Llama-3-8B)Improved
replace S-PID feedback with LQR optimal control on the same setpoint (ablation)LFS setpoint tracking · RealToxicityPromptsperplexity / toxicity tradeoffImproved
steer with A-LQR for toxicity on small modelsA-LQR · RealToxicityPrompts (Gemma-2-2B)perplexity: 8.95 -> 12.26Hurt
steer with A-LQR for truthfulnessA-LQR · TruthfulQA (Llama-3-8B)T*I: 46.22 -> 63.63Improved
steer with A-LQR for truthfulness on larger modelsA-LQR · TruthfulQA (Qwen-2.5-14B)MMLU (%): 78.80 -> 71.66Hurt
add ActAdd steering for truthfulnessActAdd · TruthfulQA (Gemma-2-2B)informativeness / MMLU: 96.08 -> 86.80 Info; 51.94 -> 35.34 MMLUHurt
intervene at all token positions (A-LQR+) instead of last token only for jailbreaking (ablation)A-LQR · AdvBench (Gemma-2-9B-Instruct)ASR: 0.1442 -> 0.7692Improved
intervene at all token positions (A-LQR+) instead of last token only for jailbreaking (ablation)A-LQR · AdvBench (Qwen2.5-14B-Instruct)ASR: 0.6923 -> 0.9712Improved
intervene at all token positions for toxicity/truthfulness/concepts (ablation)A-LQR · toxicity and truthfulness tasksprimary metrics and PPL/DistNo clear effect
A-LQR last-token steering for jailbreaking vs AASA-LQR · AdvBench (Llama-3.2-3B-Instruct)ASR: 0.9135 (AAS) -> 0.7115 (A-LQR)Hurt
reuse offline-computed LQR gains across reachable activations (linear time-varying approximation)A-LQR · Gemma-2-2B tracking rolloutstracking error vs boundImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy