arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2603.17832cs.CLcs.AIcs.LG
收藏

文本到舞台:从长篇叙述中生成空间布局

Text-to-Stage: Spatial Layouts from Long-form Narratives

Rice University(里士大学);Meta FAIR;Meta Reality Labs Research(Meta现实实验室)

Jefferson Hernandez, Swarnadeep Saha, Chenxi Whitehouse, Sanjeel Parekh, Calvin Murdock, Yuliang Li, W. Owen Brimijoin, Vamsi Krishna Ithapu, Ishwarya Ananthabhotla

AI总结:

本文研究如何通过语言模型从无结构文本推断戏剧舞台布局,提出确定性评估体系和结合拒绝SFT与RL的训练方法,在多个指标上优于基线模型。

摘要:

在本工作中,我们探索语言模型从无结构文本中展现空间推理能力的可能性,模拟人类能力并自动化受益于许多下游媒体应用的过程。具体而言,我们研究叙事到戏剧的任务:从缺乏显式空间、位置或关系线索的文本中推断戏剧舞台布局(场景、说话者位置、动作和房间类型)。我们随后引入受戏剧学启发的确定性评估套件,最后提出结合使用最佳-N采样拒绝SFT与通过GRPO从可验证奖励中进行强化学习的训练和推理配方。在仅文本的古典英语文学语料库上的实验表明,在多个指标(角色归属、空间合理性、动作经济性)上优于基线模型,同时与LLM-as-a-judge和主观人类偏好一致。

英文摘要:

In this work, we probe the ability of a language model to demonstrate spatial reasoning from unstructured text, mimicking human capabilities and automating a process that benefits many downstream media applications. Concretely, we study the narrative-to-play task: inferring stage-play layouts (scenes, speaker positions, movements, and room types) from text that lacks explicit spatial, positional, or relational cues. We then introduce a dramaturgy-inspired deterministic evaluation suite and, finally, a training and inference recipe that combines rejection SFT using Best-of-N sampling with RL from verifiable rewards via GRPO. Experiments on a text-only corpus of classical English literature demonstrate improvements over vanilla models across multiple metrics (character attribution, spatial plausibility, and movement economy), as well as alignment with an LLM-as-a-judge and subjective human preferences.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑