StoryWriter: A Multi-Agent Framework for Long Story Generation

2025-06-19

AI Summary

Large language models (LLMs) face two key challenges in long-story generation: weak discourse coherence and limited narrative complexity. To address these, we propose StoryWriter—a multi-agent collaborative framework comprising an outline agent (which generates structured outlines via event-graph modeling), a planning agent (enabling chapter-adaptive dynamic planning), and a writing agent (integrating dynamic history compression with context-aware generation). Our work introduces the first end-to-end generation paradigm driven by events, adaptive to chapter structure, and featuring dynamic history compression. We further construct LongStory—the first high-quality long-story dataset (6,000 stories, average length 8,000 words)—and perform supervised fine-tuning on Llama3.1-8B and GLM4-9B, releasing StoryWriter_GLM. Experiments demonstrate significant improvements over state-of-the-art baselines in story consistency, plot richness, and controllable story length.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Humans and AI: Game Design — Procedural Content Generation & Storytelling

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Long story generation remains a challenge for existing large language models (LLMs), primarily due to two main factors: (1) discourse coherence, which requires plot consistency, logical coherence, and completeness in the long-form generation, and (2) narrative complexity, which requires an interwoven and engaging narrative. To address these challenges, we propose StoryWriter, a multi-agent story generation framework, which consists of three main modules: (1) outline agent, which generates event-based outlines containing rich event plots, character, and event-event relationships. (2) planning agent, which further details events and plans which events should be written in each chapter to maintain an interwoven and engaging story. (3) writing agent, which dynamically compresses the story history based on the current event to generate and reflect new plots, ensuring the coherence of the generated story. We conduct both human and automated evaluation, and StoryWriter significantly outperforms existing story generation baselines in both story quality and length. Furthermore, we use StoryWriter to generate a dataset, which contains about $6,000$ high-quality long stories, with an average length of $8,000$ words. We train the model Llama3.1-8B and GLM4-9B using supervised fine-tuning on LongStory and develop StoryWriter_GLM and StoryWriter_GLM, which demonstrates advanced performance in long story generation.

Problem

• Addresses long story generation challenges in LLMs
• Ensures discourse coherence and narrative complexity
• Proposes multi-agent framework for high-quality stories

Innovation

• Multi-agent framework for long story generation
• Outline, planning, and writing agents ensure coherence
• Generates high-quality dataset for fine-tuning models