Reproduce “QRAE: Query-Relevant Attention Entropy for Context Compression in DeepSearch 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

Yingjie Tan, Haiquan Qiu, Quanming Yao, Manyi Zhong, Xianzhi Yu, Yaqing Wang

QRAE helps DeepSearch agents manage retrieved snippets by scoring intact search results, keeping selected snippets in active context, and deferring others for later reconsideration. Its paired prefills avoid a generative summary pass but add evaluation costs. Across six model–benchmark settings, task success rose 2.35 percentage points as reported token use fell about 45% and latency about 30%, within retrieved-evidence tasks.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.