arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节（Veterans Day），12 日无更新
arXiv 2602.22769cs.AIcs.LG
收藏

AMA-Bench：评估智能体应用的长时记忆

AMA-Bench: Evaluating Long-Horizon Memory for Agentic Applications

UCSD（加州大学圣塔克拉拉分校）；Recursive

Yujie Zhao, Boqin Yuan, Junbo Huang, Haocheng Yuan, Zhongming Yu, Haozhou Xu, Lanxiang Hu, Abhilash Shankarampeta, Zimeng Huang, Wentao Ni, Yuandong Tian, Jishen Zhao

阅读 PDF
arXiv 页面 ↗HTML 原文 ↗

AI总结：

提出AMA-Bench基准，通过真实与合成轨迹评估LLM智能体的长时记忆，并基于因果图与工具增强检索提出AMA-Agent系统，在基准上提升11.16%准确率。

摘要：

大型语言模型（LLM）越来越多地被用作复杂、长时应用中的自主智能体，其中有效的记忆对于持续性能至关重要。然而，现有的记忆基准主要围绕对话，而真实的智能体记忆由连续的智能体-环境交互轨迹组成，包括状态、动作、观察和工具输出。为填补这一空白，我们引入了**AMA-Bench**（任意长度的智能体记忆），一个在现实智能体设置中评估长时记忆的基准。AMA-Bench结合了来自代表性应用的真实智能体轨迹与专家策划的问答，以及可扩展到任意视野的合成轨迹与基于规则的问答。我们的研究表明，现有记忆系统表现不佳，因为它们未能捕获因果和客观信息，并严重依赖有损的基于相似性的检索。我们进一步提出了**AMA-Agent**，一个基于因果图构建和工具增强检索的记忆系统。AMA-Agent在AMA-Bench上达到**57.22%**的准确率，超过最强基线**11.16%**。资源可在[https://ama-bench.github.io/](https://ama-bench.github.io/)获取。

英文摘要：

Large Language Models (LLMs) are increasingly used as autonomous agents in complex, long-horizon applications, where effective memory is critical for sustained performance. Yet existing memory benchmarks are largely dialogue-centric, while real agent memory consists of continuous agent-environment interaction trajectories composed of states, actions, observations, and tool outputs. To address this gap, we introduce **AMA-Bench** (**A**gent **M**emory with **A**ny length), a benchmark for evaluating long-horizon memory in realistic agentic settings. AMA-Bench combines real-world agent trajectories from representative applications with expert-curated QA, as well as synthetic trajectories that scale to arbitrary horizons with rule-based QA. Our study shows that existing memory systems underperform because they fail to capture causal and objective information and rely heavily on lossy similarity-based retrieval. We further propose **AMA-Agent**, a memory system based on causality-graph construction and tool-augmented retrieval. AMA-Agent achieves **57.22%** accuracy on AMA-Bench, outperforming the strongest baseline by **11.16%**. Resources are available at: [https://ama-bench.github.io/](https://ama-bench.github.io/).

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

↑