Reproduce “Steering Readerly Dynamics: Trajectory-Guided Story Generation with Large Language Models” 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
Danrui Li, Yiwei Zhao, Mubbasir Kapadia, Lingxiao Li
The authors introduce a system that guides story generation toward reader responses. A steering model predicts the next response, a writer generates segments, and a measure model selects the closest. In humor generation with GPT-5.5, Min-K MAE falls from 14.94 to 9.73, about 35% lower, for responses measured on human-written stories; this is an LLM-measured proxy, steering one response dimension at a time.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.