NextMem: Towards Latent Factual Memory for LLM-based Agents

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Zeyu, Li, Rui, Zhao, Xiaoyan, Zhang, Yang, Wang, Wenjie, Chen, Xu, Chua, Tat-Seng
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910055087996928
author Zhang, Zeyu
Li, Rui
Zhao, Xiaoyan
Zhang, Yang
Wang, Wenjie
Chen, Xu
Chua, Tat-Seng
author_facet Zhang, Zeyu
Li, Rui
Zhao, Xiaoyan
Zhang, Yang
Wang, Wenjie
Chen, Xu
Chua, Tat-Seng
contents Memory is critical for LLM-based agents to preserve past observations for future decision-making, where factual memory serves as its foundational part. However, existing approaches to constructing factual memory face several limitations. Textual methods impose heavy context and indexing burdens, while parametric methods suffer from catastrophic forgetting and high costs. To address these challenges, we introduce NextMem, a latent factual memory framework that utilizes an autoregressive autoencoder to efficiently construct latent memory while ensuring accurate reconstruction. For better optimization, we propose a two-stage training process, including autoregressive reconstruction alignment and progressive latent substitution. We also incorporate quantization to reduce storage overhead. Extensive experiments demonstrate that NextMem achieves superior performance, and excels in retrieval, robustness, and extensibility properties. We release our code and model checkpoints at https://github.com/nuster1128/NextMem.
format Preprint
id arxiv_https___arxiv_org_abs_2603_15634
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle NextMem: Towards Latent Factual Memory for LLM-based Agents
Zhang, Zeyu
Li, Rui
Zhao, Xiaoyan
Zhang, Yang
Wang, Wenjie
Chen, Xu
Chua, Tat-Seng
Artificial Intelligence
Information Retrieval
Machine Learning
Memory is critical for LLM-based agents to preserve past observations for future decision-making, where factual memory serves as its foundational part. However, existing approaches to constructing factual memory face several limitations. Textual methods impose heavy context and indexing burdens, while parametric methods suffer from catastrophic forgetting and high costs. To address these challenges, we introduce NextMem, a latent factual memory framework that utilizes an autoregressive autoencoder to efficiently construct latent memory while ensuring accurate reconstruction. For better optimization, we propose a two-stage training process, including autoregressive reconstruction alignment and progressive latent substitution. We also incorporate quantization to reduce storage overhead. Extensive experiments demonstrate that NextMem achieves superior performance, and excels in retrieval, robustness, and extensibility properties. We release our code and model checkpoints at https://github.com/nuster1128/NextMem.
title NextMem: Towards Latent Factual Memory for LLM-based Agents
topic Artificial Intelligence
Information Retrieval
Machine Learning
url https://arxiv.org/abs/2603.15634