Search papers, labs, and topics across Lattice.
This paper presents MAGNET, a multi-agent narrative generation framework that utilizes persona-grounded character agents to create coherent long-form stories by tracking a shared world state and evolving story goals. The authors introduce ATLAS, a graph-based verification method that assesses narrative consistency and detects hallucinations in generated content. Evaluation results indicate that MAGNET significantly reduces narrative inconsistencies and hallucinations by up to 50% compared to existing single-model approaches, establishing a new standard for long-form narrative generation.
MAGNET achieves a 50% reduction in hallucinations in long-form narratives by leveraging a multi-agent approach that grounds characters in a shared world state.
Although large language models (LLMs) have demonstrated impressive creative fiction generation, they struggle to maintain narrative consistency and coherent plot lines in long-form stories. In this work, we introduce a unified framework for long-form narrative generation and verification. MAGNET, a multi-agent goal-driven narrative engine for storytelling, generates stories with persona-grounded character agents that propose actions based on a shared world state and evolving story goals, while ATLAS is a graph-based pipeline that compares scene-level world representations across a generated story to detect hallucinations. By evaluating MAGNET using an LLM editor, pairwise rubric scoring, and ATLAS, we show that our framework produces coherent narratives compared to single-model prompting and IBSEN. At 100 pages, MAGNET reduced annotations and hallucinations by 41 and 50%, respectively, compared to the single model baseline and by 34 and 45%, respectively, compared to IBSEN, with pairwise rubric evaluation showing similar results. These results suggest that long-form narratives can emerge from explicit world-state tracking and goal-driven multi-agent generation, providing a foundation for controllable and structurally coherent long-form narrative generation.
