Reproduce “A State Change Is Local, Its Consequences Are Not: Dependency-Aware State Revision for Long-Term Agent Memory” 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
Yiqi Wang, Tongyuan Zhao, Jiaqi Zhang, Yiqun Duan, Zhangkai Wu, Mingkai Zheng, Taotao Cai
StateGraph tracks support dependencies among memories and checks whether they survive changed evidence before revising downstream states. It proposes candidate links from bounded cues, then counterfactually verifies them and checks for independent support. On StateChangeBench’s 10-case depth-3+ stratum, its Impact-F1 is 75.1% versus 49.1% for a CUPMem reimplementation; this is initial evidence from 50 controlled author-constructed cases.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.