Reproduce “EvoMemState: Evolving Memory State for Personal Agents” 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
Zhishang Xiang, Yunbo Tang, Zhongqiao Liu, Junkai Chen, Zerui Chen, Qingguo Hu, Yuxuan Zhang, William Grant and 2 more
EvoMemState preserves facts and their relationships, then generates request-specific guidance for using them. It retrieves connected updates, conflicts, and constraints, and trains the guidance policy using answer-quality feedback. On MemSyco-Bench, its accuracy among questions with all required evidence retrieved is 68.5%, versus A-MEM’s 37.3%; retrieval completeness is 85.6%, and systems’ evaluated subsets may differ.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.