Sparse Query Attention (SQA): A Computationally Efficient Attention Mechanism with Query Heads Reduction

• URL: http://arxiv.org/abs/2510.01817v1
• Date: Thu, 02 Oct 2025 09:01:38 GMT
• Title: Sparse Query Attention (SQA): A Computationally Efficient Attention Mechanism with Query Heads Reduction
• Authors: Adam Filipek,
• Abstract summary: This paper introduces Sparse Query Attention (SQA), a novel attention architecture that pursues an alternative and complementary optimization path.<n>It can achieve significant throughput improvements of up to 3x in computation-bound scenarios such as model pre-training, fine-tuning, and encoder-based tasks.<n>SQA was discovered serendipitously during the development of the upcoming Reactive Transformer architecture.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The Transformer architecture, underpinned by the Multi-Head Attention (MHA) mechanism, has become the de facto standard for state-of-the-art models in artificial intelligence. However, the quadratic computational complexity of MHA with respect to sequence length presents a significant barrier to scaling, particularly for applications involving long contexts. Prevailing solutions, such as Multi-Query Attention (MQA) and Grouped-Query Attention (GQA), have effectively addressed the memory bandwidth bottleneck that dominates autoregressive inference latency by sharing Key and Value projections. While highly successful, these methods do not reduce the fundamental number of floating-point operations (FLOPs) required for the attention score computation, which remains a critical bottleneck for training and full-sequence processing. This paper introduces Sparse Query Attention (SQA), a novel attention architecture that pursues an alternative and complementary optimization path. Instead of reducing Key/Value heads, SQA reduces the number of Query heads. This architectural modification directly decreases the computational complexity of the attention mechanism by a factor proportional to the reduction in query heads, thereby lowering the overall FLOPs. This work presents the theoretical foundation of SQA, its mathematical formulation, and a family of architectural variants. Empirical benchmarks on long sequences (32k-200k tokens) demonstrate that SQA can achieve significant throughput improvements of up to 3x in computation-bound scenarios such as model pre-training, fine-tuning, and encoder-based tasks, with only a minimal impact on model quality in preliminary smallscale experiments. SQA was discovered serendipitously during the development of the upcoming Reactive Transformer architecture, suggesting its potential as a powerful tool for building more efficient and scalable models

Related papers

• Explicit Multi-head Attention for Inter-head Interaction in Large Language Models [70.96854312026319]
  Multi-head Explicit Attention (MEA) is a simple yet effective attention variant that explicitly models cross-head interaction.<n>MEA shows strong robustness in pretraining, which allows the use of larger learning rates that lead to faster convergence.<n>This enables a practical key-value cache compression strategy that reduces KV-cache memory usage by 50% with negligible performance loss.
  arXiv  Detail & Related papers  (2026-01-27T13:45:03Z)
• SpecAttn: Speculating Sparse Attention [1.6921396880325779]
  We introduce SpecAttn, a novel training-free approach that seamlessly integrates with speculative decoding techniques.<n>Our key insight is to exploit the attention weights already computed by the draft model during speculative decoding to identify important tokens for the target model.<n>SpecAttn achieves over 75% reduction in key-value cache accesses with a mere 15.29% increase in perplexity on the PG-19 dataset.
  arXiv  Detail & Related papers  (2025-10-31T17:12:34Z)
• MEC-Quant: Maximum Entropy Coding for Extremely Low Bit Quantization-Aware Training [15.099918961133866]
  Quantization-Aware Training (QAT) has driven much attention to produce efficient neural networks.<n>We argue that quantization inevitably introduce biases into the learned representation, especially under the extremely low-bit setting.<n>We propose Entropy Coding Quantization (MEC-Quant), a more principled objective that explicitly optimize on the structure of the representation.
  arXiv  Detail & Related papers  (2025-09-19T01:37:02Z)
• End-to-End On-Device Quantization-Aware Training for LLMs at Inference Cost [53.25965863436039]
  Quantization-aware training (QAT) provides a more principled solution, but its reliance on backpropagation incurs prohibitive memory costs.<n>We propose ZeroQAT, a zeroth-order optimization-based QAT framework that supports both weight and activation quantization.<n>Experiments show that ZeroQAT consistently outperforms representative PTQ and QAT baselines while requiring significantly less memory.
  arXiv  Detail & Related papers  (2025-08-21T01:18:27Z)
• Hierarchical Reasoning Model [16.223136644998203]
  HRM executes sequential reasoning tasks in a single forward pass without explicit supervision of the intermediate process.<n>With only 27 million parameters, HRM achieves exceptional performance on complex reasoning tasks using only 1000 training samples.
  arXiv  Detail & Related papers  (2025-06-26T19:39:54Z)
• Mix-QSAM: Mixed-Precision Quantization of the Segment Anything Model [0.0]
  Mix-QSAM is a mixed-precision Post-Training Quantization (PTQ) framework for the Segment Anything Model (SAM)<n>We introduce a layer-wise importance score, derived using Kullback-Leibler (KL) divergence, to quantify each layer's contribution to the model's output.<n>We also introduce cross-layer synergy, a novel metric based on causal mutual information, to capture dependencies between adjacent layers.
  arXiv  Detail & Related papers  (2025-05-08T00:08:31Z)
• RSQ: Learning from Important Tokens Leads to Better Quantized LLMs [65.5558181902098]
  Layer-wise quantization is a key technique for efficiently compressing large models without expensive retraining.<n>We propose RSQ (Rotate, Scale, then Quantize), which applies rotations to the model to mitigate outliers.<n>We demonstrate that RSQ consistently outperforms baseline methods across multiple downstream tasks and three model families.
  arXiv  Detail & Related papers  (2025-03-03T18:46:33Z)
• Tensor Product Attention Is All You Need [53.69820973900921]
  Product Attention (TPA) is a novel attention mechanism that uses tensor decompositions to represent queries, keys, and values compactly.<n>TPA achieves improved model quality alongside memory efficiency.<n>Based on TPA, we introduce the Product Attention Transformer,(T6), a new model architecture for sequence modeling.
  arXiv  Detail & Related papers  (2025-01-11T03:37:10Z)
• MQRetNN: Multi-Horizon Time Series Forecasting with Retrieval Augmentation [1.8692254863855964]
  Multi-horizon probabilistic time series forecasting has wide applicability to real-world tasks such as demand forecasting. Recent work in neural time-series forecasting mainly focus on the use of Seq2Seq architectures. We consider incorporating cross-entity information to enhance model performance by adding a cross-entity attention mechanism along with a retrieval mechanism to select which entities to attend over.
  arXiv  Detail & Related papers  (2022-07-21T14:51:58Z)
• Scaling Quantum Approximate Optimization on Near-term Hardware [49.94954584453379]
  We quantify scaling of the expected resource requirements by optimized circuits for hardware architectures with varying levels of connectivity. We show the number of measurements, and hence total time to synthesizing solution, grows exponentially in problem size and problem graph degree. These problems may be alleviated by increasing hardware connectivity or by recently proposed modifications to the QAOA that achieve higher performance with fewer circuit layers.
  arXiv  Detail & Related papers  (2022-01-06T21:02:30Z)
• Transformer-based Machine Learning for Fast SAT Solvers and Logic Synthesis [63.53283025435107]
  CNF-based SAT and MaxSAT solvers are central to logic synthesis and verification systems. In this work, we propose a one-shot model derived from the Transformer architecture to solve the MaxSAT problem.
  arXiv  Detail & Related papers  (2021-07-15T04:47:35Z)
• Once Quantization-Aware Training: High Performance Extremely Low-bit Architecture Search [112.05977301976613]
  We propose to combine Network Architecture Search methods with quantization to enjoy the merits of the two sides. We first propose the joint training of architecture and quantization with a shared step size to acquire a large number of quantized models. Then a bit-inheritance scheme is introduced to transfer the quantized models to the lower bit, which further reduces the time cost and improves the quantization accuracy.
  arXiv  Detail & Related papers  (2020-10-09T03:52:16Z)

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.