# From Profiles to Steering Vectors: Global Sparse Priors and Local Semantic Calibration for Personalized Text Generation

arXiv: https://arxiv.org/abs/2607.21620 · HTML: https://mlsanity.com/paper/2607.21620/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Activation steering methods, one of 13 directions in Mechanistic interpretability.
- Builds on 21 of its 29 references that are in the map.
- First posted on arXiv 1 Jun 2026.

## Placed in

- Mechanistic interpretability / [Activation steering methods](https://mlsanity.com/mechinterp/activation-steering-methods/)

## Builds on (most cited first, 21 of 21)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Unsupervised Dense Information Retrieval with Contrastive Learning](https://mlsanity.com/paper/2112.09118/) (2021)
- [Steering Llama 2 via Contrastive Activation Addition](https://mlsanity.com/paper/2312.06681/) (2023)
- [Steering Language Models With Activation Engineering](https://mlsanity.com/paper/2308.10248/) (2023)
- [A Survey on In-context Learning](https://mlsanity.com/paper/2301.00234/) (2022)
- [LaMP: When Large Language Models Meet Personalization](https://mlsanity.com/paper/2304.11406/) (2023)
- [TruthX: Alleviating Hallucinations by Editing Large Language Models in Truthful Space](https://mlsanity.com/paper/2402.17811/) (2024)
- [SAEs Are Good for Steering -- If You Select the Right Features](https://mlsanity.com/paper/2505.20063/) (2025)
- [A Survey on Sparse Autoencoders: Interpreting the Internal Mechanisms of Large Language Models](https://mlsanity.com/paper/2503.05613/) (2025)
- [Personalized Pieces: Efficient Personalized Large Language Models through Collaborative Efforts](https://mlsanity.com/paper/2406.10471/) (2024)
- [HYDRA: Model Factorization Framework for Black-Box LLM Personalization](https://mlsanity.com/paper/2406.02888/) (2024)
- [LLM-based Medical Assistant Personalization with Short- and Long-Term Memory Coordination](https://mlsanity.com/paper/2309.11696/) (2023)
- [Personalized Text Generation with Contrastive Activation Steering](https://mlsanity.com/paper/2503.05213/) (2025)
- [Steering off Course: Reliability Challenges in Steering Language Models](https://mlsanity.com/paper/2504.04635/) (2025)
- [LLMs + Persona-Plug = Personalized LLMs](https://mlsanity.com/paper/2409.11901/) (2024)
- [PRIME: Large Language Model Personalization with Cognitive Dual-Memory and Personalized Thought Process](https://mlsanity.com/paper/2507.04607/) (2025)
- [Learning to Rank for Multiple Retrieval-Augmented Models through Iterative Utility Maximization](https://mlsanity.com/paper/2410.09942/) (2024)
- [Can Large Language Models Understand Preferences in Personalized Recommendation?](https://mlsanity.com/paper/2501.13391/) (2025)
- [Persona-SQ: A Personalized Suggested Question Generation Framework For Real-world Documents](https://mlsanity.com/paper/2412.12445/) (2024)
- [Personalized Language Model Learning on Text Data Without User Identifiers](https://mlsanity.com/paper/2501.06062/) (2025)
- [Fints: Efficient Inference-Time Personalization for LLMs with Fine-Grained Instance-Tailored Steering](https://mlsanity.com/paper/2510.27206/) (2025)

## Abstract (arXiv)

Personalized text generation requires models to capture user-specific writing styles from historical data. Existing approaches based on retrieval, parameter-efficient fine-tuning, or activation steering either introduce inference and storage overhead or struggle to separate stylistic signals from semantic content. We propose GLASS, a training-free framework for personalized generation via Global-Local Activation Steering with Sparse priors. GLASS uses sparse autoencoders to extract a global user-style prior from historical responses and constructs local contrastive style vectors over clustered interaction scenarios. During inference, it jointly injects global and local vectors into different model layers, enabling context-aware personalization without retrieval or parameter updates. Experiments on LaMP and LongLaMP show that GLASS outperforms retrieval-, fine-tuning-, and steering-based baselines across ROUGE metrics and LLM-as-judge evaluations. Further analyses show that SAE-based representations are more robust to topic and length shifts, suggesting better disentanglement of stylistic information from semantic residue.