Researchers have developed A-MapReduce, a new framework inspired by the MapReduce paradigm to improve the efficiency of wide search tasks performed by large language model (LLM) based multi-agent systems. This approach recasts wide search as a horizontally structured retrieval problem, enabling parallel processing of massive retrieval targets through adaptive decomposition and structured result aggregation. Experiments show A-MapReduce achieves state-of-the-art performance on benchmarks like WideSearch and DeepWideSearch, outperforming strong baselines using OpenAI o3 or Gemini 2.5 Pro by significant margins in Item F1 scores, while also reducing running time and improving cost-effectiveness. AI
IMPACT Enhances efficiency and performance of LLM agents in large-scale retrieval tasks.
RANK_REASON The cluster describes a new research paper detailing a novel framework for LLM multi-agent systems. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
