agent/papers/items/2026-2606-11680-organize-then-retrieve-hierarchical-memory-navigation-for-efficient-agents.md

Paper: Organize then Retrieve: Hierarchical Memory Navigation for Efficient Agents

---

type: paper title: "Organize then Retrieve: Hierarchical Memory Navigation for Efficient Agents" authors: Hao-Lun Hsu, Nikki Lijing Kuang, Boyi Liu, Zhewei Yao, Yuxiong He year: 2026 venue: arXiv url: https://arxiv.org/abs/2606.11680 code_url: source: arxiv collected_at: 2026-07-08 published_at: 2026-06-10 updated_at: 2026-06-10 status: skimmed relevance: high topics:

• agent-evaluation
• computer-use
• embodied-agent
• memory
• planning
• rag
• reasoning methods:
• hierarchical-memory-workspace
• memory-skill-evolution
• rl-retrieval-agent benchmarks:
• ALFWorld
• LoCoMo
• LongMemEval models:

datasets:

• cs.AI
• cs.CL
• cs.LG related_concepts:
• memory-organization
• agentic-retrieval related_jobs:

related_experiments:

• KC-001-agent-memory-pilot related_projects:
• learning/agent-memory collection_score: 18 collection_queries: agent-memory

---

One-line Takeaway

Memory construction 和 retrieval 应分开优化：强模型负责长期组织，轻量策略负责沿层级 workspace 取回最小充分证据。

Pilot Skim

• problem: 联合优化构建与检索会让稀疏任务 reward 无法归因，也混淆两个不同时间尺度的职责。
• method: 高层 manager 组织带 provenance 的文件层级，retriever 用 Bash-like tools 导航并通过 RL 优化。
• evidence: ALFWorld strict context 下达到 56.7/73.9 SR；小型 Qwen retriever 在对话数据训练后迁移到其他域。
• boundary: weak manager 造成的组织缺陷无法由强 retriever 补回，系统依赖高能力 construction model。

Used In

• Agent Memory evidence matrix

Links

• arXiv: https://arxiv.org/abs/2606.11680