Lngram v2: Latent N-Gram Memory with Interpretable Discrete Representations

• URL: http://arxiv.org/abs/2609.03426v1
• Date: Thu, 03 Sep 2026 06:33:23 GMT
• Title: Lngram v2: Latent N-Gram Memory with Interpretable Discrete Representations
• Abstract summary: Transformers lack a native lookup mechanism, requiring repeated dense computation to recognize and reuse local static patterns.<n>Lngram v1 introduces tokenizer-independent conditional memory through discrete latent n-gram addressing.<n>Lngram v2 decouples the number of routes, memory dimension, and backbone width.
• Score: 8.050927352582827
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Transformers lack a native lookup mechanism, requiring repeated dense computation to recognize and reuse local static patterns. Lngram v1 introduces tokenizer-independent conditional memory through discrete latent n-gram addressing, but its memory capacity is coupled with the backbone width, limiting scalability due to high parameter and activation costs. We propose Lngram v2, which decouples the number of routes, memory dimension, and backbone width, and introduces a context-aware grouped-query attention readout to scale memory capacity independently. A zero-value Sink and counterfactual surrogate gradients further improve readout selectivity and routing trainability while preserving hard discrete addressing. Experiments across vision--language models (VLMs) of different scales show consistent improvements, including successful scaling to a 30B-parameter model. Compared with Lngram v1, Lngram v2 substantially reduces both total and activated memory parameters while maintaining or improving language modeling performance. Further analysis shows that its discrete IDs preserve substantial semantic structure of continuous hidden states, enabling semantic recovery from IDs alone and stable ID--semantic associations across datasets. These results establish Lngram v2 as an efficient and scalable latent conditional memory mechanism whose discrete addresses also provide a structured interface for analyzing internal model representations.

Related papers

• Dynamic Linear Attention [33.7476613423211]
  We propose DLA, a dynamic memory modeling framework for multi-state linear attention.<n>We pre-train DLA on two different linear attention models and evaluate on 16 datasets across three categories.
  arXiv  Detail & Related papers  (2026-06-09T09:57:48Z)
• Advancing Narrative Long Video Generation via Training-Free Identity-Aware Memory [79.01059178883817]
  IAMFlow is a training-free identity-aware memory framework that explicitly models and tracks persistent entity identities.<n>A VLM asynchronously verifies and refines attributes from rendered frames, enabling explicit entity tracking in place of implicit similarity-based matching.<n>NarraStream-Bench is a benchmark for narrative streaming video generation that features 324 multi-prompt scripts spanning six dimensions.
  arXiv  Detail & Related papers  (2026-05-18T17:54:34Z)
• Persistent Visual Memory: Sustaining Perception for Deep Generation in LVLMs [66.15429821530503]
  Persistent Visual Memory is a lightweight learnable module designed to strengthen sustained, on-demand access to visual evidence.<n>Experiments on Qwen3-VL models demonstrate that PVM brings notable improvements with negligible parameter overhead.<n>In-depth analysis reveals that PVM shows improved robustness in longer generations and accelerates internal prediction convergence.
  arXiv  Detail & Related papers  (2026-05-01T17:54:37Z)
• MoVE: Mixture of Value Embeddings -- A New Axis for Scaling Parametric Memory in Autoregressive Models [0.9222161299777548]
  We introduce $textbfMoVE (Mixture of Value Embeddings)$, a mechanism that breaks the rigid structural coupling of model capacity to computational cost.<n>MoVE decouples memory from compute by introducing a global bank of learnable value embeddings shared across all attention layers.<n>We validate MoVE through strictly controlled experiments on two representative applications of autoregressive modeling: Text Generation and Image Generation.
  arXiv  Detail & Related papers  (2026-01-30T12:07:23Z)
• LINA: Linear Autoregressive Image Generative Models with Continuous Tokens [56.80443965097921]
  Autoregressive models with continuous tokens form a promising paradigm for visual generation, especially for text-to-image (T2I) synthesis.<n>We study how to design compute-efficient linear attention within this framework.<n>We present LINA, a simple and compute-efficient T2I model built entirely on linear attention, capable of generating high-fidelity 1024x1024 images from user instructions.
  arXiv  Detail & Related papers  (2026-01-30T06:44:33Z)
• CroBIM-V: Memory-Quality Controlled Remote Sensing Referring Video Object Segmentation [0.3099118620919279]
  This paper advances RS-RVOS research through dual contributions in data and methodology.<n>First, we construct RS-RVOS Bench, the first large-scale benchmark comprising 111 video sequences, about 25,000 frames, and 213,000 temporal referring annotations.<n>Second, we propose a memory-quality-aware online referring segmentation framework, termed Memory Quality Control with Segment Anything Model (MQC-SAM)
  arXiv  Detail & Related papers  (2026-01-17T14:52:46Z)
• Fast SAM2 with Text-Driven Token Pruning [52.8350457627401]
  Segment Anything Model 2 (SAM2), a vision computation model has significantly advanced in prompt-driven video object segmentation.<n>SAM2 pipelines propagate all visual tokens produced by the image encoder through downstream temporal reasoning modules, regardless of their relevance to the target object.<n>We introduce a text-guided token pruning framework that improves inference efficiency by selectively reducing token density prior to temporal propagation.
  arXiv  Detail & Related papers  (2025-12-24T18:59:05Z)
• Dynamic Memory-enhanced Transformer for Hyperspectral Image Classification [3.5093938502961763]
  Hyperspectral image (HSI) classification remains a challenging task due to the intricate spatial-spectral correlations.<n>Existing transformer models excel in capturing long-range dependencies but often suffer from information redundancy and attention inefficiencies.<n>MemFormer introduces a memory-enhanced multi-head attention mechanism that iteratively refines a dynamic memory module.<n>A dynamic memory enrichment strategy progressively captures complex spatial and spectral dependencies, leading to more expressive feature representations.
  arXiv  Detail & Related papers  (2025-04-17T17:43:34Z)
• TF-SASM: Training-free Spatial-aware Sparse Memory for Multi-object Tracking [6.91631684487121]
  Multi-object tracking (MOT) in computer vision remains a significant challenge, requiring precise localization and continuous tracking of multiple objects in video sequences. We propose a novel memory-based approach that selectively stores critical features based on object motion and overlapping awareness. Our approach significantly improves over MOTRv2 in the DanceTrack test set, demonstrating a gain of 2.0% AssA score and 2.1% in IDF1 score.
  arXiv  Detail & Related papers  (2024-07-05T07:55:19Z)
• Self-Gated Memory Recurrent Network for Efficient Scalable HDR Deghosting [59.04604001936661]
  We propose a novel recurrent network-based HDR deghosting method for fusing arbitrary length dynamic sequences. We introduce a new recurrent cell architecture, namely Self-Gated Memory (SGM) cell, that outperforms the standard LSTM cell. The proposed approach achieves state-of-the-art performance compared to existing HDR deghosting methods quantitatively across three publicly available datasets.
  arXiv  Detail & Related papers  (2021-12-24T12:36:33Z)
• Improve Variational Autoencoder for Text Generationwith Discrete Latent Bottleneck [52.08901549360262]
  Variational autoencoders (VAEs) are essential tools in end-to-end representation learning. VAEs tend to ignore latent variables with a strong auto-regressive decoder. We propose a principled approach to enforce an implicit latent feature matching in a more compact latent space.
  arXiv  Detail & Related papers  (2020-04-22T14:41:37Z)

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.