Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2601.01885 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917449573597184 |
|---|---|
| author | Yu, Yi Yao, Liuyi Xie, Yuexiang Tan, Qingquan Feng, Jiaqi Li, Yaliang Wu, Libing |
| author_facet | Yu, Yi Yao, Liuyi Xie, Yuexiang Tan, Qingquan Feng, Jiaqi Li, Yaliang Wu, Libing |
| contents | Large language model (LLM) agents face fundamental limitations in long-horizon reasoning due to finite context windows, making effective memory management critical. Existing methods typically handle long-term memory (LTM) and short-term memory (STM) as separate components, relying on heuristics or auxiliary controllers, which limits adaptability and end-to-end optimization. In this paper, we propose Agentic Memory (AgeMem), a unified framework that integrates LTM and STM management directly into the agent's policy. AgeMem exposes memory operations as tool-based actions, enabling the LLM agent to autonomously decide what and when to store, retrieve, update, summarize, or discard information. To train such unified behaviors, we propose a three-stage progressive reinforcement learning strategy and design a step-wise GRPO to address sparse and discontinuous rewards induced by memory operations. Experiments on five long-horizon benchmarks demonstrate that AgeMem consistently outperforms strong memory-augmented baselines across multiple LLM backbones, achieving improved task performance, higher-quality long-term memory, and more efficient context usage. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_01885 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Agentic Memory: Learning Unified Long-Term and Short-Term Memory Management for Large Language Model Agents Yu, Yi Yao, Liuyi Xie, Yuexiang Tan, Qingquan Feng, Jiaqi Li, Yaliang Wu, Libing Computation and Language Large language model (LLM) agents face fundamental limitations in long-horizon reasoning due to finite context windows, making effective memory management critical. Existing methods typically handle long-term memory (LTM) and short-term memory (STM) as separate components, relying on heuristics or auxiliary controllers, which limits adaptability and end-to-end optimization. In this paper, we propose Agentic Memory (AgeMem), a unified framework that integrates LTM and STM management directly into the agent's policy. AgeMem exposes memory operations as tool-based actions, enabling the LLM agent to autonomously decide what and when to store, retrieve, update, summarize, or discard information. To train such unified behaviors, we propose a three-stage progressive reinforcement learning strategy and design a step-wise GRPO to address sparse and discontinuous rewards induced by memory operations. Experiments on five long-horizon benchmarks demonstrate that AgeMem consistently outperforms strong memory-augmented baselines across multiple LLM backbones, achieving improved task performance, higher-quality long-term memory, and more efficient context usage. |
| title | Agentic Memory: Learning Unified Long-Term and Short-Term Memory Management for Large Language Model Agents |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2601.01885 |