cs.CLNov 2, 2026

SYNC: A Training-Free Inference Controller for Dynamic Layer Selection in Multi-Persona Alignment

Authors: Yosef Nuraga Wicaksana, Muhammad Reza Qorib, Fadhil Muhammad, Alfan Farizki Wicaksono

Organizations: University of Indonesia · Carnegie Mellon University

Abstract

Steering vectors align a model at inference time without touching its weights, but eliciting several personas at once is a different problem from eliciting one. Prior multi-persona steering approaches predetermine both how persona vectors are combined and the specific layers at which they are injected. Moreover, this static commitment introduces substantial offline search costs to identify suitable target layers, despite the fact that optimal intervention strengths vary dynamically across both decoding steps and distinct personas. We introduce SYNC (SensitivitY-guided iNference Control), a novel training-free inference control framework compatible with pre-existing steering vectors. SYNC dynamically estimates each layer's sensitivity to every persona via cosine similarity trajectories, assigning each persona to its optimal injection layer at every decoding step. Across two models, two persona profiles and steering vectors from both unsupervised and supervised extraction, SYNC leads on 21 of 28 persona configurations, significantly on 15. Code is available at https://github.com/yosefnuragaw/sync.

Figures & tables

Appendix figures & tables5 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. Ready2Blend: From Natural-Language Instructions to Composable Alignment Prompts

    Sep 30, 2026Jeesu Jung, Hwan Chang, Juseon Do +5Continual Learning for LLMsLLM Alignment

  2. Inference-time Alignment via Sparse Junction Steering

    Jan 30, 2026Runyi Hu, Jie Zhang, Shiqian Zhao +7Inference-Time Language Model AlignmentLanguage Model Steering