Adapting large language models to an individual author's style from a few examples is challenging, and scientific writing sharpens the difficulty: formal conventions leave little surface variation, and authors write about their own topics, so extracted ``style'' easily entangles with content. We study style-conditioned abstract generation from a few example abstracts per author and propose three methods: (1) contrastive activation steering, (2) a network that predicts steering vectors, and (3) a hypernetwork that predicts LoRA adapters. We find a consistent trade-off between style imitation and output quality: fine-tuning buys most of the available style signal but forfeits fluency, while the hypernetwork achieves the best trade-off on both seen and unseen authors. Our steering operates at author level, contrasting an author's abstracts against style-neutral generations for the same content. This holds topic fixed, removes the need for a predefined style inventory, and outperforms inventory-based steering. % [EDIT 1a] softened "no single optimal axis" claim Moreover, our analyses demonstrate that manually extracted and predicted steering vectors are near-orthogonal yet score comparably, indicating that style conditioning here can admit at least two unrelated directions rather than requiring one particular axis.
Figures & tables
Figure 1: Both style control networks differ only in their output head. The author-specific example abstracts Ra are embedded into za , concatenated with a layer embedding (for the hypernetwork, also a module embedding) and passed through the MLP backbone. The steering head emits a vector added to the residual stream. The LoRA head emits adapter weights inserted into the attention projections. The network is queried once per target layer l∈L (and per module m for LoRA prediction).
Pred. Positive
Pred. Negative
Total
GT Positive
1,216
0, 16
1,232
GT Negative
0,00 2
1,230
1,232
Total
1,218
1,246
2,464
Table 1: Confusion matrix for the topic preservation judge. It attains 99.3% accuracy, misclassifying 18 of 2,464 abstract-introduction pairs. Negative pairs are constructed by randomly pairing abstracts with unrelated introductions.
Content
Style
Style Acc. (%)
Target author’s papers
(1)
target
human (ground truth)
53.2
(2)
target
LoRA (target author)
49.8
(3)
target
none (zero-shot)
31.0
Different author’s papers
(4)
foreign
LoRA (target author)
0 0.3
Table 2: Evaluation of the k -NN author classifier. Each row is a set of abstracts labeled with the target author, varying the paper ( Content ) and the source of writing style ( Style ). Removing style at fixed content (row 3) costs 22.2 points, yet the remaining 31.0% is attainable from content alone. Conversely, styling a different author’s paper (rows 4-5) falls to chance.
Table 3: Style and quality results on two tests sets with (a) seen and (b) unseen authors.
Figure 2: Style-quality trade-off by plotting the preference rate against style accuracy for each method, on seen (a) and unseen authors (b). The hypernetwork sits immediately before quality degrades.
L2 Norm
L1 / L2
Cosine Similarity
Layer
Manual
Pred.
Hidden
Manual
Pred.
Hidden
Manual vs. Pred.
18
13.4
14.3
39.9
17.3
40.1
11.8
−0.03
19
13.6
0 8.2
46.4
18.1
39.5
12.1
−0.05
20
15.1
12.2
49.4
19.4
39.7
12.0
−0.04
Table 4: Per-layer statistics for manual and predicted steering vectors, averaged over the 203 validation samples. Hidden indicates the unsteered activations the vectors are added to. L1/L2 measures how broadly a vector spreads its mass, with 40.4 expected of a dense Gaussian vector in 2,560 dimensions and 50.6 of a uniform one. The final column reports the average cosine similarity between the manual and predicted vector at each layer.
Layers
Style Acc.
MRR
Pref. Rate
(%)
(%)
Manual steering (validation)
18–20, ours
36.0
–
–
20–22, ours
35.0
–
–
all 36, ours
0 2.5
–
–
18–20, Konen et al. (2024)
34.0
–
–
Table 5: Effect of layer selection. Manual steering is reported on the validation set at λ=1.0 , where the configuration search measured style accuracy only (hence “ − ” for MRR and Pref. Rate). The learned interventions are reported on the test set.
Appendix figures & tables7 assets
Supplementary material from the paper’s appendix.
Appendix
Figure 3: The prompt template used to query GPT-OSS-20B to classify the topic preservation of an abstract. {abstract} is replaced with the summary of the paper. {introduction} is replaced with the introduction the abstract is compared with.
Figure 4: The prompt template used to query GPT-OSS-20B for the quality win rate evaluation. {abstract_a} is replaced with the generated abstract and {abstract_b} with the human-written ground-truth abstract. Each pair is evaluated twice with the input order swapped to eliminate positional bias.
Figure 5: The prompt template used to query Qwen3-4B-Instruct-2507 4-bit for chunk summarizations. {text_chunk} is replaced with a chunk of the paper.
Figure 6: The prompt template used to query Qwen3-4B-Instruct-2507 4-bit for papers that exceed the context window and are first split into chunks. {summary_x} is replaced with the summary of chunk x . The [...] notation is a schematic abbreviation; the actual prompt contains one numbered section block per chunk.
Figure 7: The prompt template used to query Qwen3-4B-Instruct-2507 4-bit for papers that fit within the model’s context window. {content} is replaced with the plain text of the paper.
Figure 8: The prompt template used for zero-shot, manual steering, learned steering, and hypernetwork setup. {summary} is replaced with the oueline of the paper.
Figure 9: The prompt template used for the few-shot and LoRA. {summary} is replaced with the outline of the paper. {example_1} , {example_2} and {example_3} are replaced with the example abstracts of the author.
Activation steering has shown promise for controlling LLM generation along well-defined attributes, but it remains unclear whether it can handle the multidimensional and hard-to-define nature of authorship style. We ask whether structured contrastive prompting along rhetorically-motivated dimensions can construct rich style representations directly in activation space, bypassing the need for natural language style descriptors or dedicated training. We find that the resulting directions share a common authorship backbone while conflicting on aspect-specific residuals that carry genuine stylistic signal, explaining why naive aggregation fails. We operationalize this in Aspect-Aware Activation Steering (A3S), a training-free framework that merges per-aspect contrastive directions with interference-aware aggregation and tunes steering strength per instance. A3S improves authorship style transfer where it is genuinely multi-aspect, outperforms a trained baseline in preference evaluations on out-of-domain benchmarks, and keeps target-exemplar overlap consistently low.
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 first train individual, style-specific LoRA adapters on a small set of high-resource authors: this allows for the rapid training of specialized adaptation models for each new target using layer-wise adapter mixing via reinforcement learning, necessitating only a handful of target-style training examples. AuthorMix ranks first on the combined style-meaning score among all baselines, including GPT-5.1, and substantially improves meaning preservation over the trained baselines; under human evaluation it is the only method best-or-tied on every dimension.
Sarubi Thillainathan, Ji-Ung Lee, Michael Sullivan +1
Department of Language Science and Technology, Saarland Informatics Campus Saarland University, Saarbrücken, Germany
Text style transfer (TST) is naturally a supervised task - rewrite a sentence in a target style while preserving its meaning - yet the parallel corpora that supervision requires exist for only a handful of style domains. A common workaround is to normalize an input into a style-agnostic intermediate and then stylize it into the target style, but the normalizer is typically a lightweight, task-specific paraphraser applied only at test time, feeding a correspondingly small stylizer. We observe that a style-stripping normalizer already exists at scale: neural MT systems trained on hundreds of millions of general-domain sentence pairs preserve content while regressing toward generic phrasing, so roundtrip translation through a pivot language strips stylistic signal without any task-specific training. This turns normalization from an inference-time patch into a data-generation tool. Roundtrip-translating a monolingual in-style corpus yields a pseudo-parallel corpus on which we LoRA-finetune an instruction-tuned LLM as the stylizer (RT-SFT); applying the same normalizer to test queries keeps that stylizer in-distribution. We show that across four style domains, RT-SFT outperforms state-of-the-art methods, such as few-shot in-context learning, by considerable margins. We also report on effective retrieval augmentation methods for expert style domains with strict terminology and naming conventions.