Memory is Reconstructed, Not Retrieved: Graph Memory for LLM Agents

Hugging Face Daily Papers Papers
llm-agents memory graph reconstruction long-context reasoning active-reconstruction

Summary

MRAgent introduces a novel graph-based memory framework that dynamically reconstructs memory during reasoning, achieving up to 23% improvement on long-horizon benchmarks while reducing computational costs.

Despite recent progress, LLM agents still struggle with reasoning over long interaction histories. While current memory-augmented agents rely on a static retrieve-then-reason paradigm, this rigid pipeline design prevents them from dynamically adapting memory access to intermediate evidence discovered during inference. To bridge this gap, we propose MRAgent, a framework that combines an associative memory graph with an active reconstruction mechanism. We represent memory as a Cue-Tag-Content graph, where associative tags serve as semantic bridges connecting fine-grained cues to memory contents. Operating on this structure, our active reconstruction mechanism integrates LLM reasoning directly into memory access, allowing the agent to iteratively explore and prune retrieval paths based on accumulated evidence. This ensures that memory retrieval is dynamically adapted to the reasoning context while avoiding combinatorial explosion caused by unconstrained expansion. Experiments on the LoCoMo benchmark and LongMemEval benchmark demonstrate significant improvements over strong baselines (up to 23%), while substantially reducing token and runtime cost, highlighting the effectiveness of active and associative reconstruction for long-horizon memory reasoning.
Original Article
View Cached Full Text

Cached at: 06/15/26, 09:05 AM

Paper page - Memory is Reconstructed, Not Retrieved: Graph Memory for LLM Agents

Source: https://huggingface.co/papers/2606.06036

Abstract

MRAgent combines associative memory graphs with active reconstruction to enable dynamic memory access during reasoning, improving long-horizon memory reasoning while reducing computational costs.

Despite recent progress,LLM agentsstill struggle with reasoning over long interaction histories. While currentmemory-augmented agentsrely on a staticretrieve-then-reason paradigm, this rigid pipeline design prevents them from dynamically adapting memory access to intermediate evidence discovered during inference. To bridge this gap, we propose MRAgent, a framework that combines anassociative memory graphwith anactive reconstruction mechanism. We represent memory as aCue-Tag-Content graph, where associative tags serve assemantic bridgesconnecting fine-grained cues to memory contents. Operating on this structure, ouractive reconstruction mechanismintegrates LLM reasoning directly into memory access, allowing the agent to iteratively explore and prune retrieval paths based on accumulated evidence. This ensures that memory retrieval is dynamically adapted to the reasoning context while avoidingcombinatorial explosioncaused by unconstrained expansion. Experiments on theLoCoMo benchmarkandLongMemEval benchmarkdemonstrate significant improvements over strong baselines (up to 23%), while substantially reducing token and runtime cost, highlighting the effectiveness of active and associative reconstruction for long-horizon memory reasoning.

View arXiv pageView PDFGitHub23Add to collection

Get this paper in your agent:

hf papers read 2606\.06036

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2606.06036 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2606.06036 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2606.06036 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles