UltraMemV2: Memory Networks Scaling to 120B Parameters with Superior Long-Context Learning

• URL: http://arxiv.org/abs/2508.18756v1
• Date: Tue, 26 Aug 2025 07:33:11 GMT
• Title: UltraMemV2: Memory Networks Scaling to 120B Parameters with Superior Long-Context Learning
• Authors: Zihao Huang, Yu Bao, Qiyang Min, Siyan Chen, Ran Guo, Hongzhi Huang, Defa Zhu, Yutao Zeng, Banggu Wu, Xun Zhou, Siyuan Qiao,
• Abstract summary: Memory-layer architectures offer an appealing alternative with very few memory access.<n>We present UltraMemV2, a redesigned memory-layer architecture that closes this performance gap.<n>We demonstrate that UltraMemV2 performance parity with 8-expert MoE models under same computation and parameters but significantly low memory access.
• Score: 22.029614513198663
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: While Mixture of Experts (MoE) models achieve remarkable efficiency by activating only subsets of parameters, they suffer from high memory access costs during inference. Memory-layer architectures offer an appealing alternative with very few memory access, but previous attempts like UltraMem have only matched the performance of 2-expert MoE models, falling significantly short of state-of-the-art 8-expert configurations. We present UltraMemV2, a redesigned memory-layer architecture that closes this performance gap. Our approach introduces five key improvements: integrating memory layers into every transformer block, simplifying value expansion with single linear projections, adopting FFN-based value processing from PEER, implementing principled parameter initialization, and rebalancing memory-to-FFN computation ratios. Through extensive evaluation, we demonstrate that UltraMemV2 achieves performance parity with 8-expert MoE models under same computation and parameters but significantly low memory access. Notably, UltraMemV2 shows superior performance on memory-intensive tasks, with improvements of +1.6 points on long-context memorization, +6.2 points on multi-round memorization, and +7.9 points on in-context learning. We validate our approach at scale with models up to 2.5B activated parameters from 120B total parameters, and establish that activation density has greater impact on performance than total sparse parameter count. Our work brings memory-layer architectures to performance parity with state-of-the-art MoE models, presenting a compelling alternative for efficient sparse computation.

Related papers

• AllMem: A Memory-centric Recipe for Efficient Long-context Modeling [32.025154452526856]
  Large Language Models (LLMs) encounter significant performance bottlenecks in long-sequence tasks.<n>We introduce textscAllMem, a novel and efficient hybrid architecture that integrates Sliding Window Attention (SWA) with non-linear Test-Time Training (TTT) memory networks.
  arXiv  Detail & Related papers  (2026-02-14T09:04:28Z)
• Efficient Multi-Instance Generation with Janus-Pro-Dirven Prompt Parsing [53.295515505026096]
  Janus-Pro-driven Prompt Parsing is a prompt- parsing module that bridges text understanding and layout generation.<n>MIGLoRA is a parameter-efficient plug-in integrating Low-Rank Adaptation into UNet (SD1.5) and DiT (SD3) backbones.<n>The proposed method achieves state-of-the-art performance on COCO and LVIS benchmarks while maintaining parameter efficiency.
  arXiv  Detail & Related papers  (2025-03-27T00:59:14Z)
• S2A: A Unified Framework for Parameter and Memory Efficient Transfer Learning [8.602744958104969]
  We propose a new PETL framework, called Structure to Activation (S2A), to reduce the memory footprint of activation during fine-tuning.<n>Specifically, our framework consists of: 1) Activation modules design(i.e., bias, prompt and side modules) in the parametric model structure, which results in a significant reduction of adjustable parameters and activation memory.<n>We show that our methods not only outperform existing PETL techniques, achieving a fourfold reduction in GPU memory footprint on average, but also shows competitive performance in accuracy with fewer tunable parameters.
  arXiv  Detail & Related papers  (2025-03-11T08:10:03Z)
• LM2: Large Memory Models [11.320069795732058]
  This paper introduces the Large Memory Model (LM2), a decoder-only Transformer architecture enhanced with an auxiliary memory module.<n> Experimental results on the BABILong benchmark demonstrate that the LM2model outperforms both the memory-augmented RMT model by 37.1% and the baseline Llama-3.2 model by 86.3% on average across tasks.
  arXiv  Detail & Related papers  (2025-02-09T22:11:42Z)
• Memory Layers at Scale [67.00854080570979]
  This work takes memory layers beyond proof-of-concept, proving their utility at contemporary scale.<n>On downstream tasks, language models augmented with our improved memory layer outperform dense models with more than twice the budget, as well as mixture-of-expert models when matched for both compute and parameters.<n>We provide a fully parallelizable memory layer implementation, demonstrating scaling laws with up to 128B memory parameters, pretrained to 1 trillion tokens, comparing to base models with up to 8B parameters.
  arXiv  Detail & Related papers  (2024-12-12T23:56:57Z)
• Memory-Efficient Training for Deep Speaker Embedding Learning in Speaker Verification [50.596077598766975]
  We explore a memory-efficient training strategy for deep speaker embedding learning in resource-constrained scenarios.<n>For activations, we design two types of reversible neural networks which eliminate the need to store intermediate activations.<n>For states, we introduce a dynamic quantization approach that replaces the original 32-bit floating-point values with a dynamic tree-based 8-bit data type.
  arXiv  Detail & Related papers  (2024-12-02T06:57:46Z)
• Ultra-Sparse Memory Network [8.927205198458994]
  This work introduces UltraMem, incorporating large-scale, ultra-sparse memory layer to address these limitations.<n>Our approach significantly reduces inference latency while maintaining model performance.<n>In experiments, the largest UltraMem we train has 20 million memory slots.
  arXiv  Detail & Related papers  (2024-11-19T09:24:34Z)
• MAMBA: Multi-level Aggregation via Memory Bank for Video Object Detection [35.16197118579414]
  We propose a multi-level aggregation architecture via memory bank called MAMBA. Specifically, our memory bank employs two novel operations to eliminate the disadvantages of existing methods. Compared with existing state-of-the-art methods, our method achieves superior performance in terms of both speed and accuracy.
  arXiv  Detail & Related papers  (2024-01-18T12:13:06Z)
• AdaLomo: Low-memory Optimization with Adaptive Learning Rate [59.64965955386855]
  We introduce low-memory optimization with adaptive learning rate (AdaLomo) for large language models. AdaLomo results on par with AdamW, while significantly reducing memory requirements, thereby lowering the hardware barrier to training large language models.
  arXiv  Detail & Related papers  (2023-10-16T09:04:28Z)
• UniPT: Universal Parallel Tuning for Transfer Learning with Efficient Parameter and Memory [69.33445217944029]
  PETL is an effective strategy for adapting pre-trained models to downstream domains. Recent PETL works focus on the more valuable memory-efficient characteristic. We propose a new memory-efficient PETL strategy, Universal Parallel Tuning (UniPT)
  arXiv  Detail & Related papers  (2023-08-28T05:38:43Z)
• A Model or 603 Exemplars: Towards Memory-Efficient Class-Incremental Learning [56.450090618578]
  Class-Incremental Learning (CIL) aims to train a model with limited memory size to meet this requirement. We show that when counting the model size into the total budget and comparing methods with aligned memory size, saving models do not consistently work. We propose a simple yet effective baseline, denoted as MEMO for Memory-efficient Expandable MOdel.
  arXiv  Detail & Related papers  (2022-05-26T08:24:01Z)
• Memformer: A Memory-Augmented Transformer for Sequence Modeling [55.780849185884996]
  We present Memformer, an efficient neural network for sequence modeling. Our model achieves linear time complexity and constant memory space complexity when processing long sequences.
  arXiv  Detail & Related papers  (2020-10-14T09:03:36Z)

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.