AutoMem: Automated Learning of Memory as a Cognitive Skill

  • 类型:arxiv
  • 标识:2607.01224
  • 链接:https://arxiv.org/abs/2607.01224
  • 主分类:rag
  • 形态:method
  • 被引:1
  • 被引来源:Semantic Scholar
  • S2被引:1
  • OpenAlex被引:0
  • 影响力被引:0
  • TLDR:Opting memory alone--without modifying the model's task-action behavior--improved the base agent's performance ~2x-4x, bringing a 32B open-weight model competitive with frontier systems such as Claude Opus 4.5 and Gemini 3.1 Pro Thinking.
  • OpenAlex ID:W7167055528
  • OpenAlex DOI:10.48550/arxiv.2607.01224
  • DOI:10.48550/arxiv.2607.01224
  • DOI来源:OpenAlex
  • 开放获取:green
  • 开放获取链接:https://doi.org/10.48550/arxiv.2607.01224
  • OpenAlex更新:2026-07-19
  • 待LLM分类:否
  • 标题中文:AutoMem:将记忆作为认知技能的自动化学习
  • TLDR中文:仅启用记忆(不修改模型的任务-动作行为)即可将基础 Agent 性能提升约 2–4 倍,使 32B 开源权重模型具备与 Claude Opus 4.5、Gemini 3.1 Pro Thinking 等前沿系统相竞争的能力。
  • 来源文件
  • /inbox/tom/_candidates/2026-07-06-agent-memory-tool-use-candidates.json
  • /inbox/tom/_candidates/2026-07-05-agent-rag-longcontext-candidates.json
  • /inbox/tom/_candidates/2026-07-04-agent-rag-longcontext-candidates.json
  • [S2 enrich]
  • [OpenAlex backfill]