MetaMem: Evolving Meta-Memory for Knowledge Utilization through Self-Reflective Symbolic Optimization

TL;DR

MetaMem introduces a self-evolving meta-memory framework that enhances LLMs' ability to utilize knowledge effectively by self-reflective optimization, improving reasoning and memory coherence.

Contribution

The paper presents MetaMem, a novel meta-memory system that self-optimizes through reasoning reflection, enabling better knowledge utilization in LLMs.

Findings

MetaMem outperforms baselines by over 3.6% in knowledge utilization tasks.

Self-reflective meta-memory improves reasoning coherence and memory integration.

Extensive experiments validate the effectiveness of the proposed framework.

Abstract

Existing memory systems enable Large Language Models (LLMs) to support long-horizon human-LLM interactions by persisting historical interactions beyond limited context windows. However, while recent approaches have succeeded in constructing effective memories, they often disrupt the inherent logical and temporal relationships within interaction sessions, resulting in fragmented memory units and degraded reasoning performance. In this paper, we propose MetaMem, a novel framework that augments memory systems with a self-evolving meta-memory, aiming to teach LLMs how to effectively utilize memorized knowledge. During meta-memory optimization, MetaMem iteratively distills transferable knowledge utilization experiences across different tasks by self-reflecting on reasoning processes and performing actions to update the current meta-memory state. The accumulated meta-memory units serve as explicit knowledge utilization experiences, guiding the LLM to systematically identify and integrate critical evidence from scattered memory fragments. Extensive experiments demonstrate the effectiveness of MetaMem, which significantly outperforms strong baselines by over 3.6%. All codes and datasets are available at https://github.com/OpenBMB/MetaMem.