REMORY: Learning Residual Memory for Context Compaction
Summary
REMORY introduces a neural memory network that supplements textual summaries with a bounded sequence of soft memory tokens for context compaction in long-horizon agents, enabling a frozen LLM to approach full-context performance using only 5.2% of input positions with consistent gains on Qwen3.8-27B and GLM-5.3-Flash.
View Cached Full Text
Cached at: 10/09/26, 10:20 AM
Paper page - REMORY: Learning Residual Memory for Context Compaction
Source: https://huggingface.co/papers/2610.11287
Abstract
Long-horizonagentscompacttheirhistorytocontinuewithinafinitecontextwindow,butatextualsummaryalonemaynotsupporteverysubsequentdecision.WeintroduceREMORY,aneuralmemorynetworkthatsupplementsthesummarywithaboundedsequenceofsoftmemorytokens.Giventhehistoryandsummary,thenetworklearnstogeneratetokensthathelpafrozenLLMapproximatethecontinuationitwouldproducewiththefullhistory.Thetokensareconditionedonthesummaryandappendedafterit,formingananalogueofaresidualconnectionalongthesequencedimension.OnSummHay,REMORYimprovessourceattributionatnearlyunchangedinsightcoverageandapproachesthefull-contextjointscoreusingonly5.2%oftheinputpositions.Acrosslong-horizonagentbenchmarks,Qwen3.8-27BandGLM-5.3-Flashshowconsistentgainswithresidualmemory.BothmodelsalsoexhibitsubstantiallyfewerrepeatedtooloutputsandtoolerrorsonBrowseCompandTerminal-Bench2.1.
View arXiv pageView PDFProject pageGitHub0Add to collection
Models citing this paper2
#### mocoV3/Qwen3.8-27B-REMORY-1.9B 2B• Updatedabout 4 hours ago • 49
#### mocoV3/GLM-5.3-Flash-REMORY-1.24B 1B• Updatedabout 1 hour ago
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2610.11287 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2610.11287 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
REMORY: Learning Residual Memory for Context Compaction
Remory 提出一种神经记忆网络，用有限的软记忆 token 补充文本摘要，使冻结的 LLM 在仅用 5.2% 输入位置的情况下逼近全上下文表现，并在 BrowseComp 和 Terminal-Bench 2.1 等长时程 agent 基准上带来一致提升、显著减少重复工具输出和错误。
RecMem: Recurrence-based Memory Consolidation for Efficient and Effective Long-Running LLM Agents
RecMem is a recurrence-based memory consolidation method for long-running LLM agents that reduces token consumption by up to 87% while improving accuracy, by only invoking LLMs when semantically similar interactions recur.
Less Context, More Accuracy: A Bi-Temporal Memory Engine for LLM Agents Where a Lean Retrieved Context Beats the Full History
This paper introduces Engram, an open-source bi-temporal memory engine for LLM agents that retrieves a compact context slice (∼9.6k tokens) to outperform the full-history baseline (79k tokens) by 10.4 accuracy points on LongMemEval, using a hybrid read path fusing dense, lexical, graph, and temporal signals.
Retrieval-Driven Memory Reconsolidation for Long-Term LLM Agents
The paper introduces REALM, a framework for long-term memory in LLM agents that uses retrieval-driven reconsolidation to autonomously organize memories into a cognitive graph, achieving improved performance on long-term memory benchmarks.
Understanding Is Done Early: A Depth Division of Labor in Large Language Models and Its Use for Unbounded-Context Memory
This paper introduces CoMem, a method that exploits the depth-wise division of labor in LLMs to cache intermediate residual tensors and recompute only upper layers for retrieval, enabling bounded read compute and memory independent of stored-context length. Evaluated on Qwen3-8B, CoMem achieves strong long-context performance with significant memory savings and prefill speedups.