Self-Tuning Sparse Attention: Multi-Fidelity Hyperparameter Optimization for Transformer Acceleration

• URL: http://arxiv.org/abs/2603.18417v1
• Date: Thu, 19 Mar 2026 02:24:38 GMT
• Title: Self-Tuning Sparse Attention: Multi-Fidelity Hyperparameter Optimization for Transformer Acceleration
• Abstract summary: AFBS-BO is a fully automated framework that discovers optimal layer- and head-specific hyper parameters without human intervention.<n>On Llama-2-7B, AFBS-BO accelerates hyper parameter discovery by 3.4x with 8.8x fewer evaluations than grid search.<n>By transforming sparse attention from a manually tuned primitive into a self-optimizing primitive, AFBS-BO enables plug-and-play acceleration across diverse transformer architectures and domains.
• Score: 0.0
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Sparse attention mechanisms promise to break the quadratic bottleneck of long-context transformers, yet production adoption remains limited by a critical usability gap: optimal hyperparameters vary substantially across layers and models, and current methods (e.g., SpargeAttn) rely on manual grid search to identify them. We propose AFBS-BO (Adaptive Fidelity Binary Search with Bayesian Optimization), a fully automated framework that discovers optimal layer- and head-specific hyperparameters without human intervention. Our hybrid algorithm combines Bayesian Optimization for global exploration with binary search for local refinement, leveraging multi-fidelity evaluation across sequence lengths to reduce tuning cost. On Llama-2-7B, AFBS-BO accelerates hyperparameter discovery by 3.4x with 8.8x fewer evaluations than grid search, and identifies high-sparsity configurations that outperform existing sparse attention baselines while closely matching dense attention quality. By transforming sparse attention from a manually tuned heuristic into a self-optimizing primitive, AFBS-BO enables plug-and-play acceleration across diverse transformer architectures and domains.

Related papers

• Universally Empowering Zeroth-Order Optimization via Adaptive Layer-wise Sampling [43.822941944402544]
  Zeroth-Order optimization presents a promising memory-efficient paradigm for fine-tuning Large Language Models.<n>However, its practical adoption is severely constrained by slow wall-clock convergence and high estimation variance.<n>We propose AdaLeZO, an Adaptive Layer-wise ZO optimization framework.
  arXiv  Detail & Related papers  (2026-04-20T13:37:31Z)
• AdaFuse: Accelerating Dynamic Adapter Inference via Token-Level Pre-Gating and Fused Kernel Optimization [84.25316984309725]
  Integration of dynamic, sparse structures with parameter-efficient adapters (e.g., LoRA) is a powerful technique for enhancing Large Language Models (LLMs)<n>Despite minimal increases in computational load, the computation latency often skyrockets, leading to decoding speeds slowing by over 2.5 times.<n>AdaFuse is a framework built on a tight co-design between the algorithm and the underlying hardware system to enable efficient dynamic adapter execution.
  arXiv  Detail & Related papers  (2026-03-12T12:46:42Z)
• AutoQRA: Joint Optimization of Mixed-Precision Quantization and Low-rank Adapters for Efficient LLM Fine-Tuning [23.59600455731982]
  We propose a joint optimization framework that simultaneously optimize the bit-width and LoRA rank configuration for each layer during the mixed quantized fine-tuning process.<n>Experiments show that AutoQRA achieves performance close to full-precision fine-tuning with a memory footprint comparable to uniform 4-bit methods.
  arXiv  Detail & Related papers  (2026-02-25T07:18:08Z)
• Domain Knowledge Guided Bayesian Optimization For Autonomous Alignment Of Complex Scientific Instruments [0.44654729785785796]
  We propose a domain knowledge guided Bayesian Optimization approach, which leverages physical insight to simplify the search problem.<n>We demonstrate this approach's efficacy on a challenging 12-dimensional, 6-crystal Split-and-Delay optical system.<n>Our results demonstrate a generalizable paradigm: leveraging physical insight to transform high-dimensional, coupled optimization problems into simpler representations can enable rapid and robust automated tuning.
  arXiv  Detail & Related papers  (2026-02-11T09:15:20Z)
• Nemotron-Flash: Towards Latency-Optimal Hybrid Small Language Models [97.55009021098554]
  This work aims to identify the key determinants of SLMs' real-device latency and offer generalizable principles and methodologies for SLM design and training.<n>We introduce a new family of hybrid SLMs, called Nemotron-Flash, which significantly advances the accuracy-efficiency frontier of state-of-the-art SLMs.
  arXiv  Detail & Related papers  (2025-11-24T08:46:36Z)
• Relation-Aware Bayesian Optimization of DBMS Configurations Guided by Affinity Scores [2.474203056060563]
  Database Management Systems (DBMSs) are fundamental for managing large-scale and heterogeneous data, and their performance is critically influenced by configuration parameters.<n>Recent research has focused on automated configuration optimization using machine learning; however, existing approaches still exhibit several key limitations.<n>We propose RelTune, a novel framework that represents parameter dependencies as a Graph and learns GNN-based latent embeddings that encode performancerelevant semantics.
  arXiv  Detail & Related papers  (2025-10-31T03:46:42Z)
• Memory Enhanced Fractional-Order Dung Beetle Optimization for Photovoltaic Parameter Identification [8.924286864388922]
  This paper proposes a Memory Enhanced Fractional-Order Dung Beetle Optimization (MFO-DBO) algorithm that integrates three coordinated strategies.<n>It consistently outperforms advanced DBO variants, FO-based algorithms, enhanced classical algorithms, and recent metaheuristics in terms of accuracy, robustness, convergence speed.
  arXiv  Detail & Related papers  (2025-08-09T05:48:30Z)
• Continual Adaptation: Environment-Conditional Parameter Generation for Object Detection in Dynamic Scenarios [54.58186816693791]
  environments constantly change over time and space, posing significant challenges for object detectors trained based on a closed-set assumption.<n>We propose a new mechanism, converting the fine-tuning process to a specific- parameter generation.<n>In particular, we first design a dual-path LoRA-based domain-aware adapter that disentangles features into domain-invariant and domain-specific components.
  arXiv  Detail & Related papers  (2025-06-30T17:14:12Z)
• Sensitivity-Aware Visual Parameter-Efficient Fine-Tuning [91.5113227694443]
  We propose a novel visual. sensuous-aware fine-Tuning (SPT) scheme. SPT allocates trainable parameters to task-specific important positions. Experiments on a wide range of downstream recognition tasks show that our SPT is complementary to the existing PEFT methods.
  arXiv  Detail & Related papers  (2023-03-15T12:34:24Z)
• AUTOMATA: Gradient Based Data Subset Selection for Compute-Efficient Hyper-parameter Tuning [72.54359545547904]
  We propose a gradient-based subset selection framework for hyper- parameter tuning. We show that using gradient-based data subsets for hyper- parameter tuning achieves significantly faster turnaround times and speedups of 3$times$-30$times$.
  arXiv  Detail & Related papers  (2022-03-15T19:25:01Z)
• Pre-trained Gaussian Processes for Bayesian Optimization [24.730678780782647]
  We propose a new pre-training based BO framework named HyperBO. We show bounded posterior predictions and near-zero regrets for HyperBO without assuming the "ground truth" GP prior is known.
  arXiv  Detail & Related papers  (2021-09-16T20:46:26Z)
• Amortized Auto-Tuning: Cost-Efficient Transfer Optimization for Hyperparameter Recommendation [83.85021205445662]
  We propose an instantiation--amortized auto-tuning (AT2) to speed up tuning of machine learning models. We conduct a thorough analysis of the multi-task multi-fidelity Bayesian optimization framework, which leads to the best instantiation--amortized auto-tuning (AT2)
  arXiv  Detail & Related papers  (2021-06-17T00:01:18Z)

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.