# 斯坦福 AutoMem：记忆管理成为可训练技能，32B 模型性能媲美顶尖闭源模型

- 来源：elvis (@omarsar0)
- 发布时间：2026-07-03 00:19
- AIHOT 分数：67
- AIHOT 链接：https://aihot.virxact.com/items/cmr3qcc4l013psl7lmbvi8xd4
- 原文链接：https://x.com/omarsar0/status/2072716688483831885

## AI 摘要

斯坦福大学提出 AutoMem，将智能体的记忆管理从固定模块变为可训练技能。模型自主决定编码内容、检索时机以及笔记组织方式，文件系统操作升级为一级动作。AutoMem 采用双循环机制：强 LLM 审查完整轨迹并重写记忆结构（提示词、模式、动作词表）；同时利用智能体自身良好的记忆决策作为训练信号。仅优化记忆（不改任务动作），便在 Crafter、MiniHack、NetHack 上取得 2–4 倍提升，使 32B 开放模型性能媲美 Claude Opus 4.5 和 Gemini 3.1 Pro Thinking。论文：arxiv.org/abs/2607.01224。

## 正文

// AutoMem //

I quite like this idea of metamemory.

（bookmark it）

This new research from Stanford treats agent's memory management as a trainable skill instead of a fixed module.

The model decides what to encode， when to retrieve， and how to organize its own notes， with file-system operations promoted to first-class actions right alongside task actions.

AutoMem automates this on two loops. A strong LLM reviews full trajectories and rewrites the memory structure （prompts， schemas， action vocabulary）. Then the agent's own good memory decisions across episodes become training signal to sharpen its proficiency.

Optimizing memory alone， without touching task-action behavior， lifts the base agent 2x to 4x on Crafter， MiniHack， and NetHack. That is enough to make a 32B open model competitive with Claude Opus 4.5 and Gemini 3.1 Pro Thinking. For long-horizon agents， memory is a high-leverage objective you can train for on its own.

Paper： https://arxiv.org/abs/2607.01224

Learn to build effective AI agents in our academy： https://academy.dair.ai/