dorsal/arxiv
View SchemaAtomMem : Learnable Dynamic Agentic Memory with Atomic Memory Operation
| Authors | Yupeng Huo, Yaxi Lu, Zhong Zhang, Haotian Chen, Yankai Lin |
|---|---|
| Categories | |
| ArXiv ID | 2601.08323vv2 |
| URL | https://arxiv.org/abs/2601.08323 |
| License | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ |
Abstract
Equipping agents with memory is essential for solving real-world long-horizon problems. However, most existing agent memory mechanisms rely on static and hand-crafted workflows. This limits the performance and generalization ability of these memory designs, which highlights the need for a more flexible, learning-based memory framework. In this paper, we propose AtomMem, which reframes memory management as a dynamic decision-making problem. We deconstruct high-level memory processes into fundamental atomic CRUD (Create, Read, Update, Delete) operations, transforming the memory workflow into a learnable decision process. By combining supervised fine-tuning with reinforcement learning, AtomMem learns an autonomous, task-aligned policy to orchestrate memory behaviors tailored to specific task demands. Experimental results across 3 long-context benchmarks demonstrate that the trained AtomMem-8B consistently outperforms prior static-workflow memory methods. Further analysis of training dynamics shows that our learning-based formulation enables the agent to discover structured, task-aligned memory management strategies, highlighting a key advantage over predefined routines.
{
"annotation_id": "dc0bf765-fdd3-479b-8663-6d6d17425bf7",
"date_created": "2026-02-17T05:53:15.163000Z",
"date_modified": "2026-02-17T05:53:15.163000Z",
"file_hash": "aae11faaceb19216932c689109e3581e5aa6aeb7a15617e51961fb00b8494958",
"private": false,
"record": {
"abstract": "Equipping agents with memory is essential for solving real-world long-horizon problems. However, most existing agent memory mechanisms rely on static and hand-crafted workflows. This limits the performance and generalization ability of these memory designs, which highlights the need for a more flexible, learning-based memory framework. In this paper, we propose AtomMem, which reframes memory management as a dynamic decision-making problem. We deconstruct high-level memory processes into fundamental atomic CRUD (Create, Read, Update, Delete) operations, transforming the memory workflow into a learnable decision process. By combining supervised fine-tuning with reinforcement learning, AtomMem learns an autonomous, task-aligned policy to orchestrate memory behaviors tailored to specific task demands. Experimental results across 3 long-context benchmarks demonstrate that the trained AtomMem-8B consistently outperforms prior static-workflow memory methods. Further analysis of training dynamics shows that our learning-based formulation enables the agent to discover structured, task-aligned memory management strategies, highlighting a key advantage over predefined routines.",
"arxiv_id": "2601.08323",
"authors": [
"Yupeng Huo",
"Yaxi Lu",
"Zhong Zhang",
"Haotian Chen",
"Yankai Lin"
],
"categories": [
"cs.AI"
],
"license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
"title": "AtomMem : Learnable Dynamic Agentic Memory with Atomic Memory Operation",
"url": "https://arxiv.org/abs/2601.08323",
"version": "v2"
},
"schema_id": "dorsal/arxiv",
"source": {
"execution_id": "19cc58b1-89e5-4a28-a21c-095ac1ab0676",
"id": "arXiv Dataset",
"type": "Model",
"variant": "snapshot-2026-01-17",
"version": "0.1.0"
},
"user_id": 1000002
}