Reproduce “Remember by Asking: Retrieval-Induced Memory Evolution for LLM 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
Wanqi Zhou, Jiawei Lu, Yang Wang, Zhaolong Xing, Zhen Chen, Ai Han, Haoyue Shi
RIME retrieves evidence while forming memory and answering questions. Instead of compressing sessions at once, it uses six questions to find turns, restores nearby context, and reconciles evidence with older memories. At answer time, it checks memory first and retrieves dialogue only after `NONE`. On 1,540 non-adversarial LoCoMo questions, it leads quality metrics but not token counts.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.