Implicit Style Conditioning: A Structured Style-Rewrite Framework for Low-Resource Character Modeling

• URL: http://arxiv.org/abs/2603.05933v1
• Date: Fri, 06 Mar 2026 06:04:47 GMT
• Title: Implicit Style Conditioning: A Structured Style-Rewrite Framework for Low-Resource Character Modeling
• Abstract summary: Small Language Models (SLMs) with highly stylized personas remains a challenge due to data scarcity and the complexity of style disentanglement.<n>We propose a Structured Style-Rewrite Framework that explicitly disentangles style into three interpretable dimensions.<n>By leveraging explicit reasoning traces during training as a strong inductive bias, our approach aligns the model's latent representations with structured style features.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large Language Models (LLMs) have demonstrated impressive capabilities in role-playing (RP); however, small Language Models (SLMs) with highly stylized personas remains a challenge due to data scarcity and the complexity of style disentanglement. Standard Supervised Fine-Tuning (SFT) often captures surface-level semantics while failing to reproduce the intricate syntactic and pragmatic nuances of a character, leading to "Out-Of-Character" (OOC) generation. To address this, we propose a Structured Style-Rewrite Framework that explicitly disentangles style into three interpretable dimensions: lexical signatures (via PMI), syntactic patterns (grounded in PCFG rules), and pragmatic style. Furthermore, we introduce an implicit style conditioning strategy via Chain-of-Thought (CoT) distillation. By leveraging explicit reasoning traces during training as a strong inductive bias, our approach aligns the model's latent representations with structured style features, enabling high-fidelity stylized generation without requiring explicit reasoning tokens during inference. Extensive experiments on a specific high-stylization domain (anime characters) demonstrate that our method enables a Qwen-1.7B model to outperform significantly larger baselines (e.g., 4B Vanilla SFT) in style consistency and semantic fidelity. Our approach offers a data-efficient paradigm for democratizing inference and deployment on consumer hardware.

Related papers

• Improving General Role-Playing Agents via Psychology-Grounded Reasoning and Role-Aware Policy Optimization [62.82109000444373]
  Psy-CoT is a chain-of-thought framework that decomposes pre-response reasoning into three role-specific steps.<n>Experiments on CoSER, CharacterBench, and CharacterEval demonstrate that Psy-CoT outperforms existing role-playing CoT methods.
  arXiv  Detail & Related papers  (2026-06-25T13:34:47Z)
• Stylized Text-to-Motion Generation via Hypernetwork-Driven Low-Rank Adaptation [15.45127792716575]
  Text-driven motion diffusion models are capable of generating realistic human motions, but text alone often struggles to express fine-level nuances of motion known as style.<n>Recent approaches have tackled this challenge by attaching a style injection mechanism to a pretrained text-driven diffusion model.<n>We propose a lightweight style conditioning framework that dynamically modulates a pretrained diffusion model through hypernetwork-generated LoRA parameters.<n> Experiments on the HumanML3D and 100STYLE datasets show state-of-the-art stylization results, while achieving improved stylization for unseen styles.
  arXiv  Detail & Related papers  (2026-05-13T10:51:54Z)
• Controlling Chat Style in Language Models via Single-Direction Editing [43.11304710234668]
  This paper tests the hypothesis that distinct stylistic attributes are encoded as linear directions in the model's activation space.<n>We present a lightweight, training-free method for precise style control.<n>Our approach supports linear style composition, enhances safety by ablating undesirable behaviors, and, as confirmed by experiments on over a dozen models, achieves high style adherence while preserving core capabilities at minimal computational cost.
  arXiv  Detail & Related papers  (2026-02-10T06:05:04Z)
• StyDeco: Unsupervised Style Transfer with Distilling Priors and Semantic Decoupling [5.12285618196312]
  StyDeco is an unsupervised framework that learns text representations specifically tailored for the style transfer task.<n>Our framework outperforms several existing approaches in both stylistic fidelity and structural preservation.
  arXiv  Detail & Related papers  (2025-08-02T06:17:23Z)
• Local Prompt Adaptation for Style-Consistent Multi-Object Generation in Diffusion Models [0.0]
  Local Prompt Adaptation (LPA) is a lightweight, training-free method that injects the prompt into content and style tokens.<n>On the T2I benchmark, LPA improves CLIP-prompt alignment over vanilla SDXL by +0.41% and over SD1.5 by +0.34%, with no diversity loss.<n>On our custom 50-prompt style-rich benchmark, LPA achieves +0.09% CLIP-prompt and +0.08% CLIP-style gains over baseline.
  arXiv  Detail & Related papers  (2025-07-27T01:32:13Z)
• HiCAST: Highly Customized Arbitrary Style Transfer with Adapter Enhanced Diffusion Models [84.12784265734238]
  The goal of Arbitrary Style Transfer (AST) is injecting the artistic features of a style reference into a given image/video. We propose HiCAST, which is capable of explicitly customizing the stylization results according to various source of semantic clues. A novel learning objective is leveraged for video diffusion model training, which significantly improve cross-frame temporal consistency.
  arXiv  Detail & Related papers  (2024-01-11T12:26:23Z)
• ParaGuide: Guided Diffusion Paraphrasers for Plug-and-Play Textual Style Transfer [57.6482608202409]
  Textual style transfer is the task of transforming stylistic properties of text while preserving meaning. We introduce a novel diffusion-based framework for general-purpose style transfer that can be flexibly adapted to arbitrary target styles. We validate the method on the Enron Email Corpus, with both human and automatic evaluations, and find that it outperforms strong baselines on formality, sentiment, and even authorship style transfer.
  arXiv  Detail & Related papers  (2023-08-29T17:36:02Z)
• GestureDiffuCLIP: Gesture Diffusion Model with CLIP Latents [3.229105662984031]
  GestureDiffuCLIP is a neural network framework for synthesizing realistic, stylized co-speech gestures with flexible style control. Our system learns a latent diffusion model to generate high-quality gestures and infuses the CLIP representations of style into the generator. Our system can be extended to allow fine-grained style control of individual body parts.
  arXiv  Detail & Related papers  (2023-03-26T03:35:46Z)
• Model Criticism for Long-Form Text Generation [113.13900836015122]
  We apply a statistical tool, model criticism in latent space, to evaluate the high-level structure of generated text. We perform experiments on three representative aspects of high-level discourse -- coherence, coreference, and topicality. We find that transformer-based language models are able to capture topical structures but have a harder time maintaining structural coherence or modeling coreference.
  arXiv  Detail & Related papers  (2022-10-16T04:35:58Z)
• GTAE: Graph-Transformer based Auto-Encoders for Linguistic-Constrained Text Style Transfer [119.70961704127157]
  Non-parallel text style transfer has attracted increasing research interests in recent years. Current approaches still lack the ability to preserve the content and even logic of original sentences. We propose a method called Graph Transformer based Auto-GTAE, which models a sentence as a linguistic graph and performs feature extraction and style transfer at the graph level.
  arXiv  Detail & Related papers  (2021-02-01T11:08:45Z)
• StyleDGPT: Stylized Response Generation with Pre-trained Language Models [39.526613595499356]
  We introduce a KL loss and a style classifier to steer response generation towards the target style in both a word-level and a sentence-level. Our model can significantly outperform state-of-the-art methods in terms of both style consistency and contextual coherence.
  arXiv  Detail & Related papers  (2020-10-06T09:29:50Z)
• Improve Variational Autoencoder for Text Generationwith Discrete Latent Bottleneck [52.08901549360262]
  Variational autoencoders (VAEs) are essential tools in end-to-end representation learning. VAEs tend to ignore latent variables with a strong auto-regressive decoder. We propose a principled approach to enforce an implicit latent feature matching in a more compact latent space.
  arXiv  Detail & Related papers  (2020-04-22T14:41:37Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.