AuthorMix: Modular Authorship Style Transfer via Layer-wise Adapter Mixing
This addresses the need for flexible and efficient style transfer in natural language processing, offering a lightweight solution for adapting to new authors with minimal data, though it is incremental in its approach.
The paper tackles the problem of authorship style transfer by proposing AuthorMix, a modular framework that uses layer-wise adapter mixing to enable rapid adaptation to new target authors with few examples, outperforming state-of-the-art baselines and GPT-5.1 for low-resource targets and improving meaning preservation.
The task of authorship style transfer involves rewriting text in the style of a target author while preserving the meaning of the original text. Existing style transfer methods train a single model on large corpora to model all target styles at once: this high-cost approach offers limited flexibility for target-specific adaptation, and often sacrifices meaning preservation for style transfer. In this paper, we propose AuthorMix: a lightweight, modular, and interpretable style transfer framework. We train individual, style-specific LoRA adapters on a small set of high-resource authors, allowing the rapid training of specialized adaptation models for each new target via learned, layer-wise adapter mixing, using only a handful of target style training examples. AuthorMix outperforms existing, SoTA style-transfer baselines -- as well as GPT-5.1 -- for low-resource targets, achieving the highest overall score and substantially improving meaning preservation.