Capturing Classic Authorial Style in Long-Form Story Generation with GRPO Fine-Tuning
This work addresses the problem of controllable style transfer in long-form generation for AI writing applications, representing an incremental improvement with a novel method for a known bottleneck.
The paper tackled the challenge of evaluating and optimizing authorial style in long-form story generation by proposing a two-stage pipeline using a style-similarity judge and GRPO fine-tuning, achieving an average style score of 0.893 across four target authors.
Evaluating and optimising authorial style in long-form story generation remains challenging because style is often assessed with ad hoc prompting and is frequently conflated with overall writing quality. We propose a two-stage pipeline. First, we train a dedicated style-similarity judge by fine-tuning a sentence-transformer with authorship-verification supervision, and calibrate its similarity outputs into a bounded $[0,1]$ reward. Second, we use this judge as the primary reward in Group Relative Policy Optimization (GRPO) to fine-tune an 8B story generator for style-conditioned writing, avoiding the accept/reject supervision required by Direct Preference Optimization (DPO). Across four target authors (Mark Twain, Jane Austen, Charles Dickens, Thomas Hardy), the GRPO-trained 8B model achieves higher style scores than open-weight baselines, with an average style score of 0.893 across authors. These results suggest that AV-calibrated reward modelling provides a practical mechanism for controllable style transfer in long-form generation under a moderate model size and training budget.