AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Long-Context Modeling with Dynamic Hierarchical Sparse Attention for Memory-Constrained LLM Inference

Siheng Xiong, Joe Zou, Faramarz Fekri, Yae Jee Cho

ICML 2026 · Architectures and efficiency › Attention and transformers · approach: Long-context KV and memory attention

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
add dynamic chunking (learned boundary predictor) over fixed chunks (ablation)DHSA · LongBench (density 12.5%), Llama-3.1-8B (4-bit)avg score: 28.0 -> 31.8Improved
add length-normalized robust chunk representation (ablation)DHSA · LongBench (density 12.5%), Llama-3.1-8B (4-bit)avg score: 30.7 -> 31.8Improved
remove both dynamic chunking and robust chunk representation (ablation)DHSA · LongBench (density 12.5%), Llama-3.1-8B (4-bit)avg score: 31.8 -> 27.9Hurt
increase token density from 6.25% to 25.0% (ablation)DHSA · LongBench, Llama-3.1-8B (4-bit)avg score: 27.7 -> 34.4Improved
use DHSA instead of Block Sparse Attention at 12.5% densityDHSA · LongBench, Llama-3.1-8B (4-bit)avg score: 27.9 -> 31.8Improved
use DHSA instead of Block Sparse AttentionDHSA · RULER 48K, Llama-3.1-8B (4-bit)avg accuracy: 43.1 -> 71.5Improved
use DHSA instead of Quest at 12.5% densityDHSA · LongBench, Llama-3.1-8B (4-bit)avg score: 30.9 -> 31.8Improved
sparse DHSA at 12.5% density versus dense attentionDHSA · LongBench, Llama-3.1-8B (4-bit)avg score: 32.7 -> 31.8No clear effect
Block Sparse Attention fixed-size chunks on needle retrievalBlock Sparse · Needle-in-a-Haystackretrieval accuracyHurt
add NMS to boundary selection (window 8) (ablation)DHSA boundary predictor · boundary inferenceperformanceImproved
fuse multiple features (diff, product, cosine) instead of cosine similarity only (ablation)DHSA boundary predictor · boundary detectionboundary stabilityImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 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