LLM-Powered Decentralized Generative Agents with Adaptive Hierarchical Knowledge Graph for Cooperative Planning

Hanqing Yang, Jingdi Chen, Marie Siew, Tania Lorido-Botran, Carlee Joe-Wong · 2025

Read LLM-Powered Decentralized Generative Agents with Adaptive Hierarchical Knowledge Graph for Cooperative Planning by Hanqing Yang, Jingdi Chen, Marie Siew, Tania Lorido-Botran, Carlee Joe-Wong (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-05453.txt.

Description

Hanqing Yang and colleagues proposed in 2025 a system for decentralized generative agents that plan together over long horizons in open worlds. Classic multi-agent learning leans on central training and fixed teamwork rules, which scale poorly and bend badly when worlds change. Their agents instead use large language models, outside knowledge, and language-based reasoning.

Memory is kept as a layered knowledge graph that mixes modes, and a structured protocol governs what agents share. In a new multi-agent crafting world, their system beat both classic and language model baselines. Teams reached goals in far fewer steps than lone agents. The authors credit adaptive memory and tidy communication for the gain.

Catalog record

CREATORHanqing Yang, Jingdi Chen, Marie Siew, Tania Lorido-Botran, Carlee Joe-Wong
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE58 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2502-05453

Browse the full library · About the Tower of Babel