LesenForschungRadarAnlageframework
Anmelden / Registrieren
← Zurück zur Lektüre
NewsThema · REMORY记忆网络Wesentlich1 Min. LesezeitDisconfirmAI Redaktion

Preprint: Residual Memory Network Helps Long-Horizon Agents Approach Full-Context Performance with 5.2% Input Positions

New neural memory network supplements summaries with soft tokens, improving source attribution and reducing tool errors while maintaining insight coverage.

2026-10-09 12:00 GMT+8

Key Finding: The REMORY preprint introduces a neural memory network that appends a bounded sequence of soft memory tokens after text summaries, enabling frozen LLMs to approach full-context joint scores using only 5.2% of input positions.

Background: Long-horizon agents typically compress history via textual summaries to fit finite context windows, but pure text summaries often fail to support all subsequent decisions, leading to information loss or hallucinations.

Result: On the SummHay benchmark, the method improved source attribution with nearly unchanged insight coverage. Across long-horizon agent benchmarks like BrowseComp and Terminal-Bench 2.1, Qwen3.8-27B and GLM-5.3-Flash showed consistent gains and substantially fewer repeated tool outputs and errors.

Limitation: Results are self-reported by the authors and have not yet been independently reproduced; validation is primarily limited to specific open-source model architectures.

Quellen

  • arxiv.org↗
Thema

REMORY记忆网络

REMORY是一种用于长程代理上下文压缩的神经记忆网络,通过在文本摘要后追加有界的软记忆令牌(soft memory tokens),模拟序列维度上的残差连接。该机制旨在解决传统文本摘要无法支持所有后续决策的问题,使冻结的大语言模型在仅使用…

Anmelden / Registrieren
LesenForschungRadarAnlageframework

2 Schritte1 BeiträgeAktualisiert 2026-10-10 15:24

Wie es dazu kam

  1. 2026-10-08Hanchen Xia等人提交arXiv预印本,正式提出REMORY架构,定义其为补充文本摘要的神经记忆网络。arxiv.org ↗
  2. 2026-10-09AI Weekly报道REMORY在SummHay基准上以5.2%输入位置逼近全上下文得分,并指出其在Qwen和GLM模型上减少工具错误。aiweekly.co ↗

Was noch ungeklärt ist

  • REMORY生成的软记忆令牌在极端长上下文场景下的具体延迟增加是多少?
  • 除了Qwen3.8-27B和GLM-5.3-Flash,该方法在其他开源基座模型上的泛化能力如何?
  • 论文中提到的“insight coverage nearly unchanged”是否有具体的量化误差范围数据公开?
Vollständiges Thema lesen →
Recherche fortsetzen →