TSUBASA: Improving Long-Horizon Personalization via Evolving Memory and Self-Learning with Context Distillation

• URL: http://arxiv.org/abs/2604.07894v1
• Date: Thu, 09 Apr 2026 07:04:19 GMT
• Title: TSUBASA: Improving Long-Horizon Personalization via Evolving Memory and Self-Learning with Context Distillation
• Abstract summary: We introduce TSUBASA, a two-pronged approach designed to improve memory writing via dynamic memory evolution, and memory reading via self-learning to internalize user experiences.<n>Extensive evaluations on long-horizon benchmarks using the Qwen-3 model family (4B to 32B) validate the effectiveness of TSUBASA, surpassing competitive memory-augmented systems that rely primarily on memory writing.
• Score: 6.898745685021223
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Personalized large language models (PLLMs) have garnered significant attention for their ability to align outputs with individual's needs and preferences. However, they still struggle with long-horizon tasks, such as tracking a user's extensive history of conversations or activities. Existing memory mechanisms often fail to capture evolving behaviors, and RAG paradigms are trapped by a quality-efficiency tradeoff. Meanwhile, parametric adaptation is bottlenecked by train-inference gap due to the scarcity of labeled data. To enhance the long-horizon capabilities of PLLMs, we introduce TSUBASA, a two-pronged approach designed to improve memory writing via dynamic memory evolution, and memory reading via self-learning with a context distillation objective to internalize user experiences. Extensive evaluations on long-horizon benchmarks using the Qwen-3 model family (4B to 32B) validate the effectiveness of TSUBASA, surpassing competitive memory-augmented systems that rely primarily on memory writing, such as Mem0 and Memory-R1. Our analyses further confirms that TSUBASA breaks the quality-efficiency barrier to achieve Pareto improvements, delivering robust, high-fidelity personalization with a reduced token budget.

Related papers

• Drive by Hindsight and Foresight: Tool-Grounded Synergistic Reasoning over Hierarchical Memory for Autonomous Driving [0.8419406375517101]
  VLMs have shown promise for autonomous driving yet still suffer from weak-temporal perception, and limited generalization.<n>Recent methods improve reasoning and decision-making through CoT explanations, retrieval-augmented generation or the static injection of tool outputs.<n>We present to our knowledge, the first synergistic framework that tightly couples hierarchical memory with proactive tool invocation in a closed loop.
  arXiv  Detail & Related papers  (2026-09-08T03:59:10Z)
• AutoMem: Automated Learning of Memory as a Cognitive Skill [63.07128029793901]
  We show that memory expertise is a learned skill.<n>We introduce AutoMem, a framework that automates the two axes of memory management.<n>Our results show that memory management is an independently learnable skill.
  arXiv  Detail & Related papers  (2026-07-01T17:57:03Z)
• MemTrain: Self-Supervised Context Memory Training [40.64801276577851]
  Existing memory-agent approaches are typically trained end-to-end with reinforcement learning on downstream tasks.<n>MemTrain is a self-supervised training framework for enhancing context-memory capability of LLM agents.<n>MemTrain consistently improves downstream memory-intensive reasoning performance across different models.
  arXiv  Detail & Related papers  (2026-06-02T05:56:24Z)
• Learning How and What to Memorize: Cognition-Inspired Two-Stage Optimization for Evolving Memory [53.8372503928207]
  Large language model (LLM) agents require long-term user memory for consistent personalization.<n>Existing memory systems mainly rely on static, hand-crafted update rules.<n>We introduce MemCoE, a cognition-inspired two-stage optimization framework that learns how memory should be organized and what information to update.
  arXiv  Detail & Related papers  (2026-05-01T14:45:20Z)
• AMemGym: Interactive Memory Benchmarking for Assistants in Long-Horizon Conversations [61.6579785305668]
  AMemGym is an interactive environment enabling on-policy evaluation and optimization for memory-driven personalization.<n>Our framework provides a scalable, diagnostically rich environment for advancing memory capabilities in conversational agents.
  arXiv  Detail & Related papers  (2026-03-02T15:15:11Z)
• Mem-T: Densifying Rewards for Long-Horizon Memory Agents [23.19373149519922]
  We introduce Mem-T, an autonomous memory agent that interfaces with a lightweight hierarchical memory database to perform dynamic updates and multi-turn retrieval over streaming inputs.<n>We also propose MoT-GRPO, a tree-guided reinforcement learning framework that transforms sparse terminal feedback into dense, step-wise supervision via memory operation tree backpropagation and hindsight credit assignment.
  arXiv  Detail & Related papers  (2026-01-30T14:23:33Z)
• MetaMem: Evolving Meta-Memory for Knowledge Utilization through Self-Reflective Symbolic Optimization [57.17751568928966]
  We propose MetaMem, a framework that augments memory systems with a self-evolving meta-memory.<n>During meta-memory optimization, MetaMem iteratively distills transferable knowledge utilization experiences across different tasks.<n>Extensive experiments demonstrate the effectiveness of MetaMem, which significantly outperforms strong baselines by over 3.6%.
  arXiv  Detail & Related papers  (2026-01-27T04:46:23Z)
• Fine-Mem: Fine-Grained Feedback Alignment for Long-Horizon Memory Management [63.48041801851891]
  Fine-Mem is a unified framework designed for fine-grained feedback alignment.<n> Experiments on Memalpha and MemoryAgentBench demonstrate that Fine-Mem consistently outperforms strong baselines.
  arXiv  Detail & Related papers  (2026-01-13T11:06:17Z)
• HiMem: Hierarchical Long-Term Memory for LLM Long-Horizon Agents [3.9396865837159822]
  HiMem is a hierarchical long-term memory framework for long-horizon dialogues.<n>It supports memory construction, retrieval, and dynamic updating during sustained interactions.<n>Results show HiMem consistently outperforms representative baselines in accuracy, consistency, and long-term reasoning.
  arXiv  Detail & Related papers  (2026-01-10T01:26:01Z)
• Dynamic Affective Memory Management for Personalized LLM Agents [1.7600011132381626]
  We propose a new memory management system for affective scenarios.<n>Our system achieves superior performance in personalization, logical coherence, and accuracy.<n>Our work offers new insights into the design of long-term memory systems.
  arXiv  Detail & Related papers  (2025-10-31T12:12:51Z)
• Memo: Training Memory-Efficient Embodied Agents with Reinforcement Learning [53.72709564555407]
  Memo is a transformer-based architecture and training recipe for reinforcement learning.<n>It incorporates the creation and retrieval of memory by interleaving periodic summarization tokens with the inputs of a model during training.<n>We demonstrate Memo's effectiveness on a gridworld meta-RL benchmark and a multi-object navigation task in photo-realistic indoor settings.
  arXiv  Detail & Related papers  (2025-10-22T16:24:47Z)
• FindingDory: A Benchmark to Evaluate Memory in Embodied Agents [49.18498389833308]
  We introduce a new benchmark for long-range embodied tasks in the Habitat simulator.<n>This benchmark evaluates memory-based capabilities across 60 tasks requiring sustained engagement and contextual awareness.
  arXiv  Detail & Related papers  (2025-06-18T17:06:28Z)
• SURGEON: Memory-Adaptive Fully Test-Time Adaptation via Dynamic Activation Sparsity [30.260783715373382]
  Test-time adaptation (TTA) has emerged to improve the performance of deep models by adapting them to unlabeled target data online.<n>Yet, the significant memory cost, particularly in resource-constrained terminals, impedes the effective deployment of most backward-propagation-based TTA methods.<n>To tackle memory constraints, we introduce SURGEON, a method that substantially reduces memory cost while preserving comparable accuracy improvements.
  arXiv  Detail & Related papers  (2025-03-26T09:27:09Z)
• Stable Hadamard Memory: Revitalizing Memory-Augmented Agents for Reinforcement Learning [64.93848182403116]
  Current deep-learning memory models struggle in reinforcement learning environments that are partially observable and long-term. We introduce the Stable Hadamard Memory, a novel memory model for reinforcement learning agents. Our approach significantly outperforms state-of-the-art memory-based methods on challenging partially observable benchmarks.
  arXiv  Detail & Related papers  (2024-10-14T03:50:17Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.