MiTA Attention: Efficient Fast-Weight Scaling via a Mixture of Top-k Activations

• URL: http://arxiv.org/abs/2602.01219v2
• Date: Tue, 03 Feb 2026 03:11:05 GMT
• Title: MiTA Attention: Efficient Fast-Weight Scaling via a Mixture of Top-k Activations
• Authors: Qishuai Wen, Zhiyuan Huang, Xianghan Meng, Wei He, Chun-Guang Li,
• Abstract summary: In Transformers, the expressive capacity of an N-width increases, but scaling its fast weights becomes expensive for extremely long sequences.<n>Recently, this fast-weight scaling perspective has motivated the Mixture-of-Experts (MoE) attention, which partitions the sequence into fast-weight experts and sparsely routes the tokens to them.<n>In this paper, we elevate this perspective to a unifying framework for a wide range of efficient attention methods by interpreting them as scaling fast weights through routing and compressor compression.
• Score: 11.032826710593632
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The attention operator in Transformers can be viewed as a two-layer fast-weight MLP, whose weights are dynamically instantiated from input tokens and whose width equals sequence length N. As the context extends, the expressive capacity of such an N-width MLP increases, but scaling its fast weights becomes prohibitively expensive for extremely long sequences. Recently, this fast-weight scaling perspective has motivated the Mixture-of-Experts (MoE) attention, which partitions the sequence into fast-weight experts and sparsely routes the tokens to them. In this paper, we elevate this perspective to a unifying framework for a wide range of efficient attention methods by interpreting them as scaling fast weights through routing and/or compression. Then we propose a compress-and-route strategy, which compresses the N-width MLP into a narrower one using a small set of landmark queries and constructs deformable experts by gathering top-k activated key-value pairs for each landmark query. We call this strategy a Mixture of Top-k Activations (MiTA), and refer to the resulting efficient mechanism as MiTA attention. Preliminary experiments on vision tasks demonstrate the promise of our MiTA attention and motivate further investigation on its optimization and broader applications in more challenging settings.

Related papers

• How Smoothing is N-simplicial Attention? [57.21791642118324]
  We introduce N-simplicial attention, going from pairwise token similarity to higher-order interactions, and adapt it for Rotary Position Embeddings (RoPE)<n>To help manage the increased complexity, we propose a cost-effective simplex selection enabling the model to focus its computation load onto the more task-sensitive interactions.
  arXiv  Detail & Related papers  (2025-12-17T17:10:57Z)
• SCOUT: Toward Sub-Quadratic Attention via Segment Compression for Optimized Utility in Transformers [15.142822497807236]
  We propose SCOUT, a hybrid architecture that compresses tokens locally within fixed-size segments and applies attention only over these compressed representations.<n>SCOUT retains much of the expressivity of full attention while substantially reducing the computational and memory cost.<n>We analyze SCOUT's computational and memory efficiency and evaluate it empirically on long-context language modeling and reasoning tasks.
  arXiv  Detail & Related papers  (2025-08-31T17:08:33Z)
• MMG-Vid: Maximizing Marginal Gains at Segment-level and Token-level for Efficient Video LLMs [67.75865317787708]
  MMG-Vid is a training-free visual token pruning framework for video understanding.<n>We show that MMG-Vid can maintain over 99.5% of the original performance, while effectively reducing 75% visual tokens.
  arXiv  Detail & Related papers  (2025-08-28T17:50:03Z)
• The New LLM Bottleneck: A Systems Perspective on Latent Attention and Mixture-of-Experts [5.10053312713569]
  This paper argues that recent architectural shifts, namely Multi-head Latent Attention (MLA) and Mixture-of-Experts (MoE), challenge the premise of specialized attention hardware.<n>The central challenge for next-generation Transformers is no longer accelerating single memory-bound layer.<n>Instead, the focus must shift to designing balanced systems with sufficient compute memory capacity, memory bandwidth, and high-bandwidth interconnects to manage the diverse demands of large-scale models.
  arXiv  Detail & Related papers  (2025-07-21T10:18:33Z)
• Polar Sparsity: High Throughput Batched LLM Inferencing with Scalable Contextual Sparsity [4.24164487223914]
  We introduce Polar Sparsity, highlighting a key shift in sparsity importance from dense to Attention layers as we scale batch size and sequence length.<n>We develop hardware-efficient, sparsity-aware kernels for selective computation and Attention, delivering up to (2.2times) end-to-end speed for models like OPT, LLaMA-2 & 3, across various batch sizes and sequence lengths without compromising accuracy.
  arXiv  Detail & Related papers  (2025-05-20T20:15:42Z)
• PowerAttention: Exponentially Scaling of Receptive Fields for Effective Sparse Attention [73.26995918610669]
  Large Language Models (LLMs) face efficiency bottlenecks due to the quadratic complexity of the attention mechanism when processing long contexts.<n>We introduce PowerAttention, a novel sparse attention design that facilitates effective and complete context extension.<n>Experiments demonstrate that PowerAttention outperforms existing static sparse attention methods by $5sim 40%$.
  arXiv  Detail & Related papers  (2025-03-05T15:24:11Z)
• A Stitch in Time Saves Nine: Small VLM is a Precise Guidance for Accelerating Large VLMs [65.00970402080351]
  A promising approach to accelerating large vision-language models (VLMs) is using partial information, such as attention maps from specific layers, to assess token importance and prune less essential tokens.<n>Our study reveals three key insights: (i) Partial attention information is insufficient for accurately identifying critical visual tokens, resulting in suboptimal performance, especially at low token retention ratios; (ii) Global attention information, such as the attention map aggregated across all layers, more effectively preserves essential tokens and maintains comparable performance under aggressive pruning; and (iii) The global attention map aggregated from a small VLM closely resembles that of a large VLM,
  arXiv  Detail & Related papers  (2024-12-04T13:56:44Z)
• Sparser is Faster and Less is More: Efficient Sparse Attention for Long-Range Transformers [58.5711048151424]
  We introduce SPARSEK Attention, a novel sparse attention mechanism designed to overcome computational and memory obstacles. Our approach integrates a scoring network and a differentiable top-k mask operator, SPARSEK, to select a constant number of KV pairs for each query. Experimental results reveal that SPARSEK Attention outperforms previous sparse attention methods.
  arXiv  Detail & Related papers  (2024-06-24T15:55:59Z)
• DeciMamba: Exploring the Length Extrapolation Potential of Mamba [89.07242846058023]
  We introduce DeciMamba, a context-extension method specifically designed for Mamba.<n>Experiments over real-world long-range NLP tasks show that DeciMamba can extrapolate to context lengths significantly longer than the ones seen during training.
  arXiv  Detail & Related papers  (2024-06-20T17:40:18Z)
• A Training-free Sub-quadratic Cost Transformer Model Serving Framework With Hierarchically Pruned Attention [43.211427581302715]
  We propose Hierarchically Pruned Attention (HiP) to increase context length in large language models.<n>HiP reduces the time complexity of the attention mechanism to $O(T log T)$ and the space complexity to $O(T)$, where $T$ is the sequence length.<n>We show that HiP significantly reduces both prefill and decoding latencies, as well as memory usage, while maintaining high-quality generation with minimal degradation.
  arXiv  Detail & Related papers  (2024-06-14T08:32:45Z)
• FAST: Factorizable Attention for Speeding up Transformers [1.3637227185793512]
  We present a linearly scaled attention mechanism that maintains the full representation of the attention matrix without compromising on sparsification. Results indicate that our attention mechanism has a robust performance and holds significant promise for diverse applications where self-attention is used.
  arXiv  Detail & Related papers  (2024-02-12T18:59:39Z)
• SCHEME: Scalable Channel Mixer for Vision Transformers [52.605868919281086]
  Vision Transformers have achieved impressive performance in many computation tasks.<n>We show that the dense connections can be replaced with a sparse block diagonal structure that supports larger expansion ratios.<n>We also propose the use of a lightweight, parameter-free, channel covariance attention mechanism as a parallel branch during training.
  arXiv  Detail & Related papers  (2023-12-01T08:22:34Z)
• Calibrating Undisciplined Over-Smoothing in Transformer for Weakly Supervised Semantic Segmentation [51.14107156747967]
  Weakly supervised semantic segmentation (WSSS) has attracted considerable attention because it requires fewer annotations than fully supervised approaches.<n>We propose an Adaptive Re-Activation Mechanism (AReAM) to control deep-level attention to undisciplined over-smoothing.<n>AReAM substantially improves segmentation performance compared with existing WSSS methods, reducing noise while sharpening focus on relevant semantic regions.
  arXiv  Detail & Related papers  (2023-05-04T19:11:33Z)
• Parameterization of Cross-Token Relations with Relative Positional Encoding for Vision MLP [52.25478388220691]
  Vision multi-layer perceptrons (MLPs) have shown promising performance in computer vision tasks. They use token-mixing layers to capture cross-token interactions, as opposed to the multi-head self-attention mechanism used by Transformers. We propose a new positional spacial gating unit (PoSGU) to efficiently encode the cross-token relations for token mixing.
  arXiv  Detail & Related papers  (2022-07-15T04:18:06Z)
• Weighted QMIX: Expanding Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning [66.94149388181343]
  We present a new version of a popular $Q$-learning algorithm for MARL. We show that it can recover the optimal policy even with access to $Q*$. We also demonstrate improved performance on predator-prey and challenging multi-agent StarCraft benchmark tasks.
  arXiv  Detail & Related papers  (2020-06-18T18:34:50Z)

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.