Trained Persistent Memory for Frozen Encoder--Decoder LLMs: Six Architectural Methods

• URL: http://arxiv.org/abs/2603.16413v1
• Date: Tue, 17 Mar 2026 11:51:21 GMT
• Title: Trained Persistent Memory for Frozen Encoder--Decoder LLMs: Six Architectural Methods
• Abstract summary: We show that persistent memory in a frozen language model is feasible -- even under severe resource constraints.<n>We implement six architectural methods spanning three injection points and four write mechanisms.<n>We argue that full end-to-end training with larger models, larger data, and orders-of-magnitude larger memory will yield substantially stronger results.
• Score: 0.0
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Frozen encoder--decoder language models are stateless: the latent representation is discarded after every forward pass, so no information persists across sessions. This paper presents a \textbf{proof-of-concept pilot study} showing that persistent memory in the \emph{continuous latent space} of a frozen LLM is feasible -- even under severe resource constraints (a single frozen Flan-T5-XL backbone, small trainable adapters, a single dataset). We implement six architectural methods spanning three injection points and four write mechanisms; unlike text-level memory systems, every write and read is a differentiable operation on dense vectors. After training only the adapter, the memory bank continues to accumulate at inference time without gradients, enabling \emph{conversational learning}. Under a forgetting-curve evaluation on LoCoMo at two capacity scales (1$\times$ and 10$\times$), the stateless baseline scores exactly zero; at 10$\times$ all six trained adapters produce positive memory-recall curves; at 1$\times$ three methods collapse, revealing capacity as a critical design parameter. Because the memory bank is a compact numerical array, it can be scaled to arbitrarily large capacity without altering the backbone. We argue that full end-to-end training with larger models, larger data, and orders-of-magnitude larger memory will yield substantially stronger results; this pilot study establishes the feasibility baseline and design-space taxonomy that such efforts require.

Related papers

• Dual-Grained Agent Memory and Shapley Context Attribution for Multimodal Agentic Learner [55.80248281761638]
  multimodal large language models (MLLMs) deliver impressive perception yet still falter on scientific and mathematical reasoning.<n>We propose DG-Mem, a dual-grained agentic memory framework that augments a frozen MLLM with a non- externally stored memory built once from training-time rollouts and consulted read-only at test time.
  arXiv  Detail & Related papers  (2026-08-24T13:56:01Z)
• The Orthogonalized Read Is a Removable Training Scaffold for Recurrent Memory [0.0]
  The effect replicates, but it is not a memory improvement.<n>Training on this task is a long chance plateau followed by a sharp escape.<n>Much of the published gain needs no architecture at all.
  arXiv  Detail & Related papers  (2026-07-02T18:54:23Z)
• 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)
• Memory Grafting: Scaling Language Model Pre-training via Offline Conditional Memory [65.39827296429527]
  Scaling conditional memory offers a promising way to increase language-model capacity.<n>Existing methods such as Engram learn large memory tables from scratch during pre-training.<n>We propose Memory Grafting, a conditional memory scaling method that utilizes frozen hidden states from a grafting model as conditional n-gram memory.
  arXiv  Detail & Related papers  (2026-05-20T09:35:04Z)
• MemGym: a Long-Horizon Memory Environment for LLM Agents [69.79226770543049]
  We present MemGym, a benchmark for agentic memory that unifies existing agent gyms and in-house memory-grounded pipelines.<n>MemGym reports memory-isolated scores that decouple memory performance from reasoning, retrieval, and tool-use ability.<n>Our synthetic pipelines for MEMGYM-CODEQA and MEMGYM-DR are length-controllable, ablation-verified at every stage, and tightly aligned with downstream scenarios.
  arXiv  Detail & Related papers  (2026-05-20T07:25:33Z)
• Trained Persistent Memory for Frozen Decoder-Only LLMs [0.0]
  We show that trained memory adapters give a frozen encoder-decoder backbone persistent latent-space memory.<n>On LoCoMo we find a striking inductive-bias dichotomy: at $1times$ capacity, three methods with strong architectural priors achieve retained-memory scores of $7-18%$.<n>These findings establish persistent latent-space memory as a general paradigm spanning major transformer families.
  arXiv  Detail & Related papers  (2026-03-20T19:11:02Z)
• NextMem: Towards Latent Factual Memory for LLM-based Agents [58.35585202907478]
  NextMem is a latent factual memory framework that utilizes an autoregressive autoencoder to efficiently construct latent memory.<n>Extensive experiments demonstrate that NextMem achieves superior performance.
  arXiv  Detail & Related papers  (2026-02-26T14:35:27Z)
• Out of the Memory Barrier: A Highly Memory Efficient Training System for LLMs with Million-Token Contexts [68.79341332280062]
  Training Large Language Models (LLMs) on long contexts is severely constrained by prohibitive GPU memory overhead, not training time.<n>We introduce OOMB, a highly memory-efficient training system that directly confronts this barrier.<n>Our approach employs a chunk-recurrent training framework with on-the-fly activation recomputation, which maintains a constant activation memory footprint.
  arXiv  Detail & Related papers  (2026-02-02T13:52:40Z)
• Memory Bank Compression for Continual Adaptation of Large Language Models [1.9336815376402718]
  Continual learning aims to update Large Language Models with new information without erasing previously acquired knowledge.<n>Memory-augmented approaches address this by equipping LLMs with a memory bank, that is an external memory module which stores information for future use.<n>We propose MBC, a model that compresses the memory bank through a codebook optimization strategy during online adaptation learning.
  arXiv  Detail & Related papers  (2026-01-02T17:22:34Z)
• Mem-α: Learning Memory Construction via Reinforcement Learning [20.916677456417464]
  Large language model (LLM) agents are constrained by limited context windows.<n>Current memory-augmented agents depend on pre-defined instructions and tools for memory updates.<n>Mem-alpha is a reinforcement learning framework that trains agents to effectively manage complex memory systems.
  arXiv  Detail & Related papers  (2025-09-30T08:02:34Z)
• Cut Your Losses in Large-Vocabulary Language Models [102.6981011879656]
  We propose Cut Cross-Entropy (CCE), a method that computes the cross-entropy loss without materializing the logits for all tokens into global memory.<n>CCE reduces the memory footprint of the loss from 24 GB to 1 MB, and the total training-time memory consumption of the head from 28 GB to 1 GB.
  arXiv  Detail & Related papers  (2024-11-13T20:30:15Z)
• BitStack: Any-Size Compression of Large Language Models in Variable Memory Environments [53.71158537264695]
  Large language models (LLMs) have revolutionized numerous applications, yet their deployment remains challenged by memory constraints on local devices.<n>We introduce textbfBitStack, a novel, training-free weight compression approach that enables megabyte-level trade-offs between memory usage and model performance.
  arXiv  Detail & Related papers  (2024-10-31T13:26:11Z)
• $\text{Memory}^3$: Language Modeling with Explicit Memory [22.572376536612015]
  We equip large language models (LLMs) with explicit memory, a memory format cheaper than model parameters and text retrieval-augmented generation (RAG) As a preliminary proof of concept, we train from scratch a 2.4B LLM, which achieves better performance than much larger LLMs and RAG models. We introduce a memory circuitry theory to support the externalization of knowledge, and present novel techniques including a memory sparsification mechanism that makes storage tractable.
  arXiv  Detail & Related papers  (2024-07-01T11:07:23Z)
• Constrained Few-shot Class-incremental Learning [14.646083882851928]
  Continually learning new classes from fresh data without forgetting previous knowledge of old classes is a very challenging research problem. We propose C-FSCIL, which is architecturally composed of a frozen meta-learned feature extractor, a trainable fixed-size fully connected layer, and a rewritable dynamically growing memory. C-FSCIL provides three update modes that offer a trade-off between accuracy and compute-memory cost of learning novel classes.
  arXiv  Detail & Related papers  (2022-03-30T18:19:36Z)
• Layered gradient accumulation and modular pipeline parallelism: fast and efficient training of large language models [0.0]
  We analyse the shortest possible training time for different configurations of distributed training. We introduce two new methods, textitlayered gradient accumulation and textitmodular pipeline parallelism, which together cut the shortest training time by half.
  arXiv  Detail & Related papers  (2021-06-04T19:21:49Z)
• TinyTL: Reduce Activations, Not Trainable Parameters for Efficient On-Device Learning [78.80707950262214]
  On-device learning enables edge devices to continually adapt the AI models to new data. Existing work solves this problem by reducing the number of trainable parameters. We present Tiny-Transfer-Learning (TinyTL) for memory-efficient on-device learning.
  arXiv  Detail & Related papers  (2020-07-22T18:39:53Z)

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.