Zhaochen Yu, Yingcheng Wu, Zhenfei Yin, Kaiyuan Chen, Zhe Zhao, Mengdi Wang, Shuicheng Yan, Ling Yang
The paper introduces Recuris, a recursive Experiential-Working Memory architecture for long-horizon agent harnesses, which improves task success by grounding skill use in current needs. Recuris enables agents to continuously transform accumulated experience into effective long-horizon behavior. This leads to significant improvements in task success across various benchmarks.
The proposed architecture has practical significance as it enables agents to perform well in long-horizon tasks, which is crucial for real-world applications. Recuris can be used to improve the performance of agents in complex tasks, leading to more effective decision-making and problem-solving. This can have a significant impact on various industries, including robotics and autonomous systems.
Thank you to arXiv for use of its open access interoperability. arxiv.org. aibrief.ai is not affiliated with or endorsed by arXiv.