SYNC: A Training-Free Inference Controller for Dynamic Layer Selection in Multi-Persona Alignment
cs.CL, cs.AI
Submitted: 2026-11-02
Updated: 2026-11-03
Code: https://github.com/yosefnuragaw/sync
Terminology
Sources
- Personalized Steering of Large Language Models: Versatile Steering Vectors Through Bi-directional Preference Optimization
- Persona Vectors: Monitoring and Controlling Character Traits in Language Models
- Scaling Instruction-Finetuned Language Models
- BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
- Where to Steer: Input-Dependent Layer Selection for Steering Improves LLM Alignment
- The Capacity for Moral Self-Correction in Large Language Models
- RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models
- The Llama 3 Herd of Models
- Word Embeddings Are Steers for Language Models
- Mistral 7B
- Programming Refusal with Conditional Activation Steering
- Steering Vector Fields for Context-Aware Inference-Time Control in Large Language Models
- gpt-oss-120b & gpt-oss-20b Model Card
- Training language models to follow instructions with human feedback
- BILLY: Steering Large Language Models via Merging Persona Vectors for Creative Generation
- The Linear Representation Hypothesis and the Geometry of Large Language Models
- Discovering Language Model Behaviors with Model-Written Evaluations
- Analyzing the Generalization and Reliability of Steering Vectors
- LLaMA: Open and Efficient Foundation Language Models
- Steering Language Models With Activation Engineering
Related papers
- ALARM: Audio-Language Alignment for Reasoning Models
- ThinkSum: Probabilistic reasoning over sets using large language models
- Coherence boosting: When your pretrained language model is not paying enough attention
- From Facts to Insights: A Persona-Driven Dual Memory Framework and Dataset for Role-Playing Agents
- Improving General Role-Playing Agents via Psychology-Grounded Reasoning and Role-Aware Policy Optimization
- The Art of Asking: Multilingual Prompt Optimization for Synthetic Data