---
title: "AutoMem: Automated Learning of Memory as a Cognitive Skill"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2607.01224v1"
permalink: "https://feed.astralor.com/items/itm_c36e5a21b5b7f24e405a75c8"
published: "2026-07-01T17:57:03.000Z"
score: 59
channel: "ai"
tags: ["AutoMem","LLM","Memory Management","Long-Horizon Tasks"]
---

# AutoMem: Automated Learning of Memory as a Cognitive Skill

## At a glance

AutoMem proposes treating memory management as a trainable skill for LLMs, automating two loops: one uses a strong LLM to review full trajectories and iteratively revise memory structures, the other extracts good memory decisions from many episodes as training signals. In three long-horizon games (Crafter, MiniHack, NetHack), optimizing memory alone improves baseline performance ~2-4x, making a 32B open-weight model competitive with Claude Opus 4.5 and Gemini 3.1 Pro Thinking. Results show memory management is an independently learnable skill with high leverage for long-horizon tasks.

## Key points

- Promotes file-system operations to first-class memory actions alongside task actions, letting the model decide memory management
- Dual-loop automation: structure loop (strong LLM reviews trajectories to improve memory structures) and proficiency loop (trains memory skill from good decisions)
- In three procedurally generated long-horizon games, optimizing memory alone yields 2-4x improvement, making 32B model competitive with frontier systems

## Full text

Memory expertise is a learned skill: knowing what to encode, when to retrieve, and how to organize knowledge--a capacity known in cognitive science as metamemory. We bring this perspective to LLMs by treating memory management as a trainable skill. We promote file-system operations to first-class memory actions alongside task actions, letting the model itself decide how to manage its memory. This memory skill improves along two axes: the structure that supports it (prompts, file schemas, action vocabulary), and the proficiency of the model exercising it. Both axes resist manual optimization: episodes in long-horizon tasks run for thousands of steps, and a single memory mistake can hide long before it surfaces, making human review of full trajectories impractical. We introduce AutoMem, a framework that automates both axes. In the first loop, a strong LLM reviews complete agent trajectories and iteratively revises the memory structure that shapes how the agent interacts with its memory files. In the second loop, the agent's own good memory decisions are identified from many episodes and used as training signal to sharpen the model's memory proficiency directly. Across three procedurally generated long-horizon games (Crafter, MiniHack, and NetHack), optimizing 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. Our results show that memory management is an independently learnable skill, and a high-leverage objective yielding large gains on long-horizon tasks.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_c36e5a21b5b7f24e405a75c8