DRESSing Up LLM: Efficient Stylized Question-Answering via Style Subspace Editing

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Ma, Xinyu, Xu, Yifeng, Lin, Yang, Wang, Tianlong, Chu, Xu, Gao, Xin, Zhao, Junfeng, Wang, Yasha
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916581778391040
author Ma, Xinyu
Xu, Yifeng
Lin, Yang
Wang, Tianlong
Chu, Xu
Gao, Xin
Zhao, Junfeng
Wang, Yasha
author_facet Ma, Xinyu
Xu, Yifeng
Lin, Yang
Wang, Tianlong
Chu, Xu
Gao, Xin
Zhao, Junfeng
Wang, Yasha
contents We introduce DRESS, a novel approach for generating stylized large language model (LLM) responses through representation editing. Existing methods like prompting and fine-tuning are either insufficient for complex style adaptation or computationally expensive, particularly in tasks like NPC creation or character role-playing. Our approach leverages the over-parameterized nature of LLMs to disentangle a style-relevant subspace within the model's representation space to conduct representation editing, ensuring a minimal impact on the original semantics. By applying adaptive editing strengths, we dynamically adjust the steering vectors in the style subspace to maintain both stylistic fidelity and semantic integrity. We develop two stylized QA benchmark datasets to validate the effectiveness of DRESS, and the results demonstrate significant improvements compared to baseline methods such as prompting and ITI. In short, DRESS is a lightweight, train-free solution for enhancing LLMs with flexible and effective style control, making it particularly useful for developing stylized conversational agents. Codes and benchmark datasets are available at https://github.com/ArthurLeoM/DRESS-LLM.
format Preprint
id arxiv_https___arxiv_org_abs_2501_14371
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle DRESSing Up LLM: Efficient Stylized Question-Answering via Style Subspace Editing
Ma, Xinyu
Xu, Yifeng
Lin, Yang
Wang, Tianlong
Chu, Xu
Gao, Xin
Zhao, Junfeng
Wang, Yasha
Computation and Language
Artificial Intelligence
Machine Learning
We introduce DRESS, a novel approach for generating stylized large language model (LLM) responses through representation editing. Existing methods like prompting and fine-tuning are either insufficient for complex style adaptation or computationally expensive, particularly in tasks like NPC creation or character role-playing. Our approach leverages the over-parameterized nature of LLMs to disentangle a style-relevant subspace within the model's representation space to conduct representation editing, ensuring a minimal impact on the original semantics. By applying adaptive editing strengths, we dynamically adjust the steering vectors in the style subspace to maintain both stylistic fidelity and semantic integrity. We develop two stylized QA benchmark datasets to validate the effectiveness of DRESS, and the results demonstrate significant improvements compared to baseline methods such as prompting and ITI. In short, DRESS is a lightweight, train-free solution for enhancing LLMs with flexible and effective style control, making it particularly useful for developing stylized conversational agents. Codes and benchmark datasets are available at https://github.com/ArthurLeoM/DRESS-LLM.
title DRESSing Up LLM: Efficient Stylized Question-Answering via Style Subspace Editing
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2501.14371