Learning to Reason for Long-Form Story Generation

• URL: http://arxiv.org/abs/2503.22828v1
• Date: Fri, 28 Mar 2025 18:48:26 GMT
• Title: Learning to Reason for Long-Form Story Generation
• Authors: Alexander Gurung, Mirella Lapata,
• Abstract summary: We propose a general story-generation task (Next-Chapter Prediction) and a reward formulation (Verified Rewards via Completion Likelihood Improvement)<n>We learn to reason over a story's condensed information and generate a detailed plan for the next chapter.<n>Our reasoning is evaluated via the chapters it helps a story-generator create, and compared against non-trained and supervised finetuning (SFT) baselines.
• Score: 98.273323001781
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Generating high-quality stories spanning thousands of tokens requires competency across a variety of skills, from tracking plot and character arcs to keeping a consistent and engaging style. Due to the difficulty of sourcing labeled datasets and precise quality measurements, most work using large language models (LLMs) for long-form story generation uses combinations of hand-designed prompting techniques to elicit author-like behavior. This is a manual process that is highly dependent on the specific story-generation task. Motivated by the recent success of applying RL with Verifiable Rewards to domains like math and coding, we propose a general story-generation task (Next-Chapter Prediction) and a reward formulation (Verified Rewards via Completion Likelihood Improvement) that allows us to use an unlabeled book dataset as a learning signal for reasoning. We learn to reason over a story's condensed information and generate a detailed plan for the next chapter. Our reasoning is evaluated via the chapters it helps a story-generator create, and compared against non-trained and supervised finetuning (SFT) baselines. Pairwise human judgments reveal the chapters our learned reasoning produces are preferred across almost all metrics, and the effect is more pronounced in Scifi and Fantasy genres.

Related papers

• StoryWeaver: A Unified World Model for Knowledge-Enhanced Story Character Customization [36.14275850149665]
  We propose a novel knowledge graph, namely Character Graph (textbfCG), which comprehensively represents various story-related knowledge.<n>We then introduce StoryWeaver, an image generator that achieve Customization via Character Graph (textbfC-CG), capable of consistent story visualization with rich text semantics.
  arXiv  Detail & Related papers  (2024-12-10T10:16:50Z)
• Agents' Room: Narrative Generation through Multi-step Collaboration [54.98886593802834]
  We propose a generation framework inspired by narrative theory that decomposes narrative writing into subtasks tackled by specialized agents. We show that Agents' Room generates stories preferred by expert evaluators over those produced by baseline systems.
  arXiv  Detail & Related papers  (2024-10-03T15:44:42Z)
• Generating Visual Stories with Grounded and Coreferent Characters [63.07511918366848]
  We present the first model capable of predicting visual stories with consistently grounded and coreferent character mentions.<n>Our model is finetuned on a new dataset which we build on top of the widely used VIST benchmark.<n>We also propose new evaluation metrics to measure the richness of characters and coreference in stories.
  arXiv  Detail & Related papers  (2024-09-20T14:56:33Z)
• DataNarrative: Automated Data-Driven Storytelling with Visualizations and Texts [27.218934418961197]
  We introduce a novel task for data story generation and a benchmark containing 1,449 stories from diverse sources. To address the challenges of crafting coherent data stories, we propose a multiagent framework employing two LLM agents. While our agentic framework generally outperforms non-agentic counterparts in both model-based and human evaluations, the results also reveal unique challenges in data story generation.
  arXiv  Detail & Related papers  (2024-08-09T21:31:33Z)
• Guiding and Diversifying LLM-Based Story Generation via Answer Set Programming [1.7889842797216124]
  Large language models (LLMs) are capable of generating stories in response to open-ended user requests. We propose using a higher-level and more abstract symbolic specification of high-level story structure to guide and diversify story generation.
  arXiv  Detail & Related papers  (2024-06-01T21:14:25Z)
• Robust Preference Learning for Storytelling via Contrastive Reinforcement Learning [53.92465205531759]
  Controlled automated story generation seeks to generate natural language stories satisfying constraints from natural language critiques or preferences. We train a contrastive bi-encoder model to align stories with human critiques, building a general purpose preference model. We further fine-tune the contrastive reward model using a prompt-learning technique to increase story generation robustness.
  arXiv  Detail & Related papers  (2022-10-14T13:21:33Z)
• Unsupervised Neural Stylistic Text Generation using Transfer learning and Adapters [66.17039929803933]
  We propose a novel transfer learning framework which updates only $0.3%$ of model parameters to learn style specific attributes for response generation. We learn style specific attributes from the PERSONALITY-CAPTIONS dataset.
  arXiv  Detail & Related papers  (2022-10-07T00:09:22Z)
• Incorporating Commonsense Knowledge into Story Ending Generation via Heterogeneous Graph Networks [16.360265861788253]
  We propose a Story Heterogeneous Graph Network (SHGN) to explicitly model both the information of story context at different levels and the multi-grained interactive relations among them. In detail, we consider commonsense knowledge, words and sentences as three types of nodes. We design two auxiliary tasks to implicitly capture the sentiment trend and key events lie in the context.
  arXiv  Detail & Related papers  (2022-01-29T09:33:11Z)
• Topic Adaptation and Prototype Encoding for Few-Shot Visual Storytelling [81.33107307509718]
  We propose a topic adaptive storyteller to model the ability of inter-topic generalization. We also propose a prototype encoding structure to model the ability of intra-topic derivation. Experimental results show that topic adaptation and prototype encoding structure mutually bring benefit to the few-shot model.
  arXiv  Detail & Related papers  (2020-08-11T03:55:11Z)

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.