Reproduce “From Role-Play to Drama-Interaction: An LLM Solution” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Weiqi Wu, Hongqiu Wu, Lai Jiang, Xingyuan Liu, Jiale Hong, Hai Zhao, Min Zhang
Researchers from Shanghai Jiao Tong University and Harbin Institute of Technology developed a comprehensive LLM-based system for interactive drama, extending capabilities beyond single-character role-play to manage entire dramatic narratives with multiple characters and plot progression. The system enables immersive and dynamic storytelling experiences where users interact with characters and environments within predefined storylines.
Key insights
- Formalizing interactive drama with a structured "drama script" format and "Narrative Chain" allows LLMs to manage complex, multi-character narratives while balancing player agency.
- An "Auto-Drama" pipeline can effectively generate diverse drama scripts from existing stories, mitigating data scarcity for training drama LLMs.
- "Sparse Instruction Tuning" significantly enhances LLMs' ability to follow complex, multi-faceted instructions required for drama orchestration.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.