Live intelligence

Latest AI updates

1 updates
Research

OpenKAGE: Open-Ended Lifelong Learning via Knowledge-Sharing Agent Groups

OpenKAGE is an agent-group evolution framework for open-ended lifelong learning in which agents autonomously explore open worlds and acquire new capabilities by accumulating and reusing knowledge from their own interactions and those of others. It combines a two-layer knowledge representation, separating situated memories from transferable experience and skills with their applicability conditions, with a knowledge-sharing mechanism that consolidates textual experience into an evolving shared knowledge graph while sharing executable skills verbatim. In Minecraft, the framework supports sustained exploration and lets agent groups reach advanced technological milestones, and the accumulated knowledge transfers to a newly initialized agent group on a different backbone model, improving survival and exploration progress.

lifelong learningmulti-agent systemsknowledge transfer