wuyang
65 lines
1.7 KiB
Markdown
65 lines
1.7 KiB
Markdown
# Paper: Infini Memory: Maintainable Topic Documents for Long-Term LLM Agent Memory
|
||
|
||
---
|
||
type: paper
|
||
title: "Infini Memory: Maintainable Topic Documents for Long-Term LLM Agent Memory"
|
||
authors: Suozhao Ji, Baodong Wu, Zehao Wang, Lei Xia, Qingping Li, Ruisong Wang, Wenbo Ding, Zhenhua Zhu, et al.
|
||
year: 2026
|
||
venue: arXiv
|
||
url: https://arxiv.org/abs/2606.10677
|
||
code_url:
|
||
source: arxiv
|
||
collected_at: 2026-07-08
|
||
published_at: 2026-06-09
|
||
updated_at: 2026-06-09
|
||
status: skimmed
|
||
relevance: high
|
||
topics:
|
||
- agent-evaluation
|
||
- memory
|
||
- rag
|
||
- tool-use
|
||
methods:
|
||
- topic-documents
|
||
- buffered-consolidation
|
||
- agentic-retrieval
|
||
benchmarks:
|
||
- MemoryAgentBench
|
||
- LongMemEval
|
||
models:
|
||
-
|
||
datasets:
|
||
- cs.AI
|
||
- cs.CL
|
||
related_concepts:
|
||
- semantic-memory
|
||
- selective-forgetting
|
||
related_jobs:
|
||
-
|
||
related_experiments:
|
||
- KC-001-agent-memory-pilot
|
||
related_projects:
|
||
- learning/agent-memory
|
||
collection_score: 18
|
||
collection_queries: agent-memory
|
||
---
|
||
|
||
## One-line Takeaway
|
||
|
||
长期事实应被维护为带证据和时间的主题文档，并允许 Agent 多步检查，而不是只保存孤立记录和一次检索结果。
|
||
|
||
## Pilot Skim
|
||
|
||
- problem: 孤立记录难以聚合证据、修订事实和维护跨 session 一致性。
|
||
- method: observation 先进入 buffer，再巩固到 topic documents；读取时迭代 search、grep 和 read_lines。
|
||
- evidence: MemoryAgentBench overall 64.7；LongMemEval 上 summary-only 41.7、summary+BM25 76.0、agentic 79.3。
|
||
- boundary: multi-hop selective forgetting 仍困难；多步 retrieval 和周期维护增加计算成本。
|
||
|
||
## Used In
|
||
|
||
- [Agent Memory evidence matrix](../../learning/agent-memory/evidence-matrix.md)
|
||
|
||
## Links
|
||
|
||
- arXiv: https://arxiv.org/abs/2606.10677
|