LoSA: Locality Aware Sparse Attention for Block-Wise Diffusion Language Models

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Xi, Haocheng, Singh, Harman, Hu, Yuezhou, Hooper, Coleman, Tiwari, Rishabh, Tomar, Aditya, Lee, Minjae, Kang, Wonjun, Mahoney, Michael, Xu, Chenfeng, Keutzer, Kurt, Gholami, Amir
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866914470654115840
author Xi, Haocheng
Singh, Harman
Hu, Yuezhou
Hooper, Coleman
Tiwari, Rishabh
Tomar, Aditya
Lee, Minjae
Kang, Wonjun
Mahoney, Michael
Xu, Chenfeng
Keutzer, Kurt
Gholami, Amir
author_facet Xi, Haocheng
Singh, Harman
Hu, Yuezhou
Hooper, Coleman
Tiwari, Rishabh
Tomar, Aditya
Lee, Minjae
Kang, Wonjun
Mahoney, Michael
Xu, Chenfeng
Keutzer, Kurt
Gholami, Amir
contents Block-wise diffusion language models (DLMs) generate multiple tokens in any order, offering a promising alternative to the autoregressive decoding pipeline. However, they still remain bottlenecked by memory-bound attention in long-context scenarios. Naive sparse attention fails on DLMs due to a KV Inflation problem, where different queries select different prefix positions, making the union of accessed KV pages large. To address this, we observe that between consecutive denoising steps, only a small fraction of active tokens exhibit significant hidden-state changes, while the majority of stable tokens remain nearly constant. Based on this insight, we propose LOSA (Locality-aware Sparse Attention), which reuses cached prefix-attention results for stable tokens and applies sparse attention only to active tokens. This substantially shrinks the number of KV indices that must be loaded, yielding both higher speedup and higher accuracy. Across multiple block-wise DLMs and benchmarks, LOSA preserves near-dense accuracy while significantly improving efficiency, achieving up to +9 points in average accuracy at aggressive sparsity levels while maintaining 1.54x lower attention density. It also achieves up to 4.14x attention speedup on RTX A6000 GPUs, demonstrating the effectiveness of the proposed method.
format Preprint
id arxiv_https___arxiv_org_abs_2604_12056
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle LoSA: Locality Aware Sparse Attention for Block-Wise Diffusion Language Models
Xi, Haocheng
Singh, Harman
Hu, Yuezhou
Hooper, Coleman
Tiwari, Rishabh
Tomar, Aditya
Lee, Minjae
Kang, Wonjun
Mahoney, Michael
Xu, Chenfeng
Keutzer, Kurt
Gholami, Amir
Computation and Language
Machine Learning
Block-wise diffusion language models (DLMs) generate multiple tokens in any order, offering a promising alternative to the autoregressive decoding pipeline. However, they still remain bottlenecked by memory-bound attention in long-context scenarios. Naive sparse attention fails on DLMs due to a KV Inflation problem, where different queries select different prefix positions, making the union of accessed KV pages large. To address this, we observe that between consecutive denoising steps, only a small fraction of active tokens exhibit significant hidden-state changes, while the majority of stable tokens remain nearly constant. Based on this insight, we propose LOSA (Locality-aware Sparse Attention), which reuses cached prefix-attention results for stable tokens and applies sparse attention only to active tokens. This substantially shrinks the number of KV indices that must be loaded, yielding both higher speedup and higher accuracy. Across multiple block-wise DLMs and benchmarks, LOSA preserves near-dense accuracy while significantly improving efficiency, achieving up to +9 points in average accuracy at aggressive sparsity levels while maintaining 1.54x lower attention density. It also achieves up to 4.14x attention speedup on RTX A6000 GPUs, demonstrating the effectiveness of the proposed method.
title LoSA: Locality Aware Sparse Attention for Block-Wise Diffusion Language Models
topic Computation and Language
Machine Learning
url https://arxiv.org/abs/2604.12056