AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

TileSparse: Arithmetic-Intensity-Aware Sparse Attention for Compute-Bound LLM Decoding

Chao Wang, Pengfei Zuo, Zhangyu Chen, Qihui Zhou, Tsung-Yi Ho, Ming-Chang Yang

ICML 2026 · Architectures and efficiency › Attention and transformers · approach: Sparse attention for long-context LLMs

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace K-only sparsity with QK 2D Sparsity (prune Q-heads per key) under the same compute budgetTileSparse on DeepSeek-V3-0324 · RULER NIAH, low budgetaccuracy: 40% accuracy gain in low-budget settingsImproved
use tiered configuration (critical tier full heads, ambient tier fewer heads) instead of single-tier QK sparsity (ablation)TileSparse · RULER, 65,536-entry budgetaccuracyImproved
tiered QK 2D sparsity instead of K-only sparsity (x=512) (ablation)TileSparse · RULER, 65,536-entry budgetaccuracyImproved
add dynamic AutoTuner over static configurations (ablation)TileSparse · RULER, 65,536-entry budgetaccuracyImproved
vary the number of critical tokens at full heads (ablation)TileSparse · RULER, 65,536-entry budgetaccuracyMixed
TileSparse instead of ArkVale to reach 99% of full-attention accuracyTileSparse · RULERcompute budget reduction: 18.3%Improved
TileSparse instead of ArkVale to reach 99% of full-attention accuracyTileSparse · LongBenchcompute budget reduction: 40.8%Improved
TileSparse instead of ArkVale attention operatorvLLM Triton MLA operator on A100 · LongBenchend-to-end attention speedup: 3.33× -> 3.99×Improved
run AutoTuner every N=10,000 stepsTileSparse · decode overheadamortized overhead: ~0.01%No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy