Scaling Creative Writing Beyond Story-Centric Data with Attribute-Guided Genre Expansion

2026-08-14

AI Summary

This study addresses the limitation of homogeneous genres in creative writing datasets that constrains model capabilities. We propose an attribute-guided genre expansion framework that decouples thematic and formal control. By integrating attribute prompting, human annotation, and quality filtering, we construct a high-quality corpus comprising 50,000 samples across 13 genres, enabling scalable generation of non-narrative data. Experimental results demonstrate that models fine-tuned on this corpus significantly outperform baselines and those trained on existing datasets on multi-genre creation benchmarks. These findings confirm that our approach effectively enhances large language models' mastery of diverse genre conventions and their generative proficiency across varied literary forms, thereby mitigating the performance bottlenecks associated with limited stylistic diversity in current training data.

Abstract

High-quality creative writing data for large language models (LLMs) remains dominated by story-centric data, limiting models' ability to follow the structural and functional conventions of diverse creative formats. We propose an attribute-guided genre expansion framework for scaling creative writing data beyond story generation. By separating thematic breadth from genre-form control, our framework leverages human-authored story prompts as diverse creative seeds, while utilizing manually curated genre attributes to enforce distinct structural, stylistic, and formatting conventions. We combine these to prompt strong LLMs for genre-faithful query-response pairs, which are then quality-filtered. Applying this framework, we construct the Multi-Genre Collection, a 50K-example corpus spanning 13 creative genres, including story, rap, lyrics, scripts, game design, character design, and other creative formats. Experiments across out-of-distribution writing benchmarks and held-out genre diagnostics demonstrate that models fine-tuned on our data consistently surpass not only base models and writing-specialized baselines, but also models trained on existing writing corpora. Genre-count ablations further indicate that controlled genre expansion, rather than story-centric scaling alone, is a key driver of robust creative writing capability.

Problem

• Creative Writing
• Story-Centric Data
• Genre Diversity
• Large Language Models
• Data Scaling

Innovation

• Attribute-Guided Genre Expansion
• Non-Story-Centric Data
• Multi-Genre Collection
• Genre-Form Control
• Creative Writing Scaling