Dynamic Mixture of Latent Memories for Self-Evolving Agents

• URL: http://arxiv.org/abs/2605.21951v1
• Date: Thu, 21 May 2026 03:35:10 GMT
• Title: Dynamic Mixture of Latent Memories for Self-Evolving Agents
• Abstract summary: MoLEM is a generative mixture of latent memory framework based on a dynamic mixture-of-experts (MoE)<n>We train the framework on continual-learning sequences spanning math, science, and code domains.<n>After the full continual-learning sequence, our method improves the average accuracy by 10.40% over the Vanilla pretrained baseline.
• Score: 57.20419419302731
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Achieving self-evolution in intelligent agents requires the continual accumulation of new knowledge across changing task sequences without forgetting previously acquired abilities. Existing approaches either internalize knowledge by updating model parameters, which induces catastrophic forgetting, or rely on external memory, which fails to genuinely enhance the model's intrinsic capabilities. We propose MoLEM, a generative mixture of latent memory framework based on a dynamic mixture-of-experts (MoE). We treat multiple experts as independent carriers to generate memory. A router selects and weights experts through key-query matching, and the aggregated latent memory is injected into the reasoning process. The base model for reasoning remains entirely frozen, with all experiential knowledge internalized into the additional modules, avoiding catastrophic forgetting. For continual learning, each training stage is paired with a lightweight autoencoder that selects the appropriate routing group at inference, and inputs that match no stage fall back to the pretrained model. Experiments train the framework on continual-learning sequences spanning math, science, and code domains. After training, we evaluate the framework on the corresponding test sets to measure task learning and competence preservation across continual adaptation stages. After the full continual-learning sequence, our method improves the average accuracy by 10.40% over the Vanilla pretrained baseline, while none of the competing methods consistently exceed this baseline across different training orders.

Related papers

• Modular Memory is the Key to Continual Learning Agents [100.09688599754465]
  We argue that combining the strengths of In-Weight Learning (IWL) and the newly emerged capabilities of In-Context Learning (ICL) through the design of modular memory is the missing piece for continual adaptation at scale.<n>We outline a conceptual framework for modular memory-centric architectures that leverage ICL for rapid adaptation and knowledge accumulation, and IWL for stable updates to model capabilities.
  arXiv  Detail & Related papers  (2026-03-02T11:40:05Z)
• Towards Continuous Intelligence Growth: Self-Training, Continual Learning, and Dual-Scale Memory in SuperIntelliAgent [10.571643330948858]
  SuperIntelliAgent is an agentic learning framework that couples a trainable small diffusion model (the learner) with a frozen large language model (the verifier)<n>Unlike conventional supervised fine-tuning, SuperIntelliAgent learns autonomously without annotation.<n>We posit that pairing a trainable learner with a reasoning-capable verifier forms a minimal reliable unit of growing intelligence.
  arXiv  Detail & Related papers  (2025-11-28T18:32:49Z)
• Beyond Memorization: Extending Reasoning Depth with Recurrence, Memory and Test-Time Compute Scaling [60.63703438729223]
  We show how different architectures and training methods affect model multi-step reasoning capabilities.<n>We confirm that increasing model depth plays a crucial role for sequential computations.
  arXiv  Detail & Related papers  (2025-08-22T18:57:08Z)
• What Do Learning Dynamics Reveal About Generalization in LLM Reasoning? [83.83230167222852]
  We find that a model's generalization behavior can be effectively characterized by a training metric we call pre-memorization train accuracy. By connecting a model's learning behavior to its generalization, pre-memorization train accuracy can guide targeted improvements to training strategies.
  arXiv  Detail & Related papers  (2024-11-12T09:52:40Z)
• Towards Robust Continual Learning with Bayesian Adaptive Moment Regularization [51.34904967046097]
  Continual learning seeks to overcome the challenge of catastrophic forgetting, where a model forgets previously learnt information. We introduce a novel prior-based method that better constrains parameter growth, reducing catastrophic forgetting. Results show that BAdam achieves state-of-the-art performance for prior-based methods on challenging single-headed class-incremental experiments.
  arXiv  Detail & Related papers  (2023-09-15T17:10:51Z)
• Complementary Learning Subnetworks for Parameter-Efficient Class-Incremental Learning [40.13416912075668]
  We propose a rehearsal-free CIL approach that learns continually via the synergy between two Complementary Learning Subnetworks. Our method achieves competitive results against state-of-the-art methods, especially in accuracy gain, memory cost, training efficiency, and task-order.
  arXiv  Detail & Related papers  (2023-06-21T01:43:25Z)
• Self-Activating Neural Ensembles for Continual Reinforcement Learning [23.00149997940467]
  Self-Activating Neural Ensembles (SANE) uses a modular architecture designed to avoid catastrophic forgetting without making any assumptions. During training, new modules are created as needed and only activated modules are updated to ensure that unused modules remain unchanged. This system enables our method to retain and leverage old skills, while growing and learning new ones.
  arXiv  Detail & Related papers  (2022-12-31T07:11:05Z)
• Dynamic Regret Analysis for Online Meta-Learning [0.0]
  The online meta-learning framework has arisen as a powerful tool for the continual lifelong learning setting. This formulation involves two levels: outer level which learns meta-learners and inner level which learns task-specific models. We establish performance in terms of dynamic regret which handles changing environments from a global prospective. We carry out our analyses in a setting, and in expectation prove a logarithmic local dynamic regret which explicitly depends on the total number of iterations.
  arXiv  Detail & Related papers  (2021-09-29T12:12:59Z)
• Neuromodulated Neural Architectures with Local Error Signals for Memory-Constrained Online Continual Learning [4.2903672492917755]
  We develop a biologically-inspired light weight neural network architecture that incorporates local learning and neuromodulation. We demonstrate the efficacy of our approach on both single task and continual learning setting.
  arXiv  Detail & Related papers  (2020-07-16T07:41:23Z)
• Automatic Recall Machines: Internal Replay, Continual Learning and the Brain [104.38824285741248]
  Replay in neural networks involves training on sequential data with memorized samples, which counteracts forgetting of previous behavior caused by non-stationarity. We present a method where these auxiliary samples are generated on the fly, given only the model that is being trained for the assessed objective. Instead the implicit memory of learned samples within the assessed model itself is exploited.
  arXiv  Detail & Related papers  (2020-06-22T15:07:06Z)

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.