Paper Detail
Minhua Lin, Zhiwei Zhang, Hanqing Lu, Hui Liu, Xianfeng Tang, Qi He, Xiang Zhang, Suhang Wang
Memory-augmented LLM agents maintain external memory banks to support long-horizon interaction, yet most existing systems treat construction, retrieval, and utilization as isolated subroutines. This creates two coupled challenges: strategic blindness on the forward path of the memory cycle, where construction and retrieval are driven by local heuristics rather than explicit strategic reasoning, and sparse, delayed supervision on the backward path, where downstream failures rarely translate into direct repairs of the memory bank. To address these challenges, we propose MemMA, a plug-and-play multi-agent framework that coordinates the memory cycle along both the forward and backward paths. On the forward path, a Meta-Thinker produces structured guidance that steers a Memory Manager during construction and directs a Query Reasoner during iterative retrieval. On the backward path, MemMA introduces in-situ self-evolving memory construction, which synthesizes probe QA pairs, verifies the current memory, and converts failures into repair actions before the memory is finalized. Extensive experiments on LoCoMo show that MemMA consistently outperforms existing baselines across multiple LLM backbones and improves three different storage backends in a plug-and-play manner. Our code is publicly available at https://github.com/ventr1c/memma.
No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.
No ranking explanation is available yet.
No tags.
@misc{lin2026memma,
title = {MemMA: Coordinating the Memory Cycle through Multi-Agent Reasoning and In-Situ Self-Evolution},
author = {Minhua Lin and Zhiwei Zhang and Hanqing Lu and Hui Liu and Xianfeng Tang and Qi He and Xiang Zhang and Suhang Wang},
year = {2026},
abstract = {Memory-augmented LLM agents maintain external memory banks to support long-horizon interaction, yet most existing systems treat construction, retrieval, and utilization as isolated subroutines. This creates two coupled challenges: strategic blindness on the forward path of the memory cycle, where construction and retrieval are driven by local heuristics rather than explicit strategic reasoning, and sparse, delayed supervision on the backward path, where downstream failures rarely translate into },
url = {https://huggingface.co/papers/2603.18718},
keywords = {memory-augmented LLM agents, external memory banks, memory cycle, Meta-Thinker, Memory Manager, Query Reasoner, in-situ self-evolving memory construction, probe QA pairs, downstream failures, repair actions, code available, huggingface daily},
eprint = {2603.18718},
archiveprefix = {arXiv},
}
{}