wuyang
1.8 KiB
1.8 KiB
Paper: Exploring Cross-Scenario Generality of Agentic Memory Systems: Diagnostics and a Strong Baseline
type: paper title: "Exploring Cross-Scenario Generality of Agentic Memory Systems: Diagnostics and a Strong Baseline" authors: Zhikai Chen, Jialiang Gu, Junyu Yin, Xianxuan Long, Shenglai Zeng, Xiaoze Liu, Kai Guo, Keren Zhou, et al. year: 2026 venue: arXiv url: https://arxiv.org/abs/2606.04315 code_url: source: arxiv collected_at: 2026-07-08 published_at: 2026-06-03 updated_at: 2026-06-03 status: skimmed relevance: high topics:
- agent-evaluation
- memory
- planning
- rag
- tool-use methods:
- agentic-memory-harness
- schema-diagnostics
- active-retrieval benchmarks:
- LoCoMo
- HotpotQA
- AMABench
- ALFWorld models:
datasets:
- cs.AI related_concepts:
- schema-commitment
- agentic-retrieval related_jobs:
related_experiments:
- KC-001-agent-memory-pilot related_projects:
- learning/agent-memory collection_score: 19 collection_queries: agent-memory
One-line Takeaway
复杂索引会在写入时承诺 schema；保留原始证据并让 Agent 按问题主动读取，往往有更好的跨场景通用性。
Pilot Skim
- problem: 为单一场景设计的 memory 在对话、QA、trajectory 和动态任务之间难以泛化。
- method: 诊断 representation loss 和 retrieval loss，并提出带主动工具循环的 AutoMEM。
- evidence: AutoMEM 在 LoCoMo 为 67.3，对 long context 的 61.5；ALFWorld 诊断显示 golden procedure 仍低于训练后的 actor。
- boundary: agentic harness 不是每个场景最省 token；动态任务上限可能在 policy 而非 memory。