AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Dynamic Linear Attention

Xin Wang, Hui Shen, Boyuan Zheng, Xueshen Liu, Minkyoung Cho, Zhongwei Wan, Zesen Zhao, Zhuoqing Mao, Shen Yan, Mi Zhang

ICML 2026 · Architectures and efficiency › Attention and transformers · approach: State space and recurrent sequence models · 0 citations (OpenAlex)

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 log-linear fixed merging with DLAMamba-2-780M · Commonsense reasoning averageaccuracy: 31.0 -> 34.8Improved
replace log-linear fixed merging with DLAGated DeltaNet-1.3B · Commonsense reasoning averageaccuracy: 32.6 -> 34.8Improved
replace log-linear with DLAMamba-2-780M · In-context retrieval averageaccuracy: 6.3 -> 13.7Improved
replace log-linear with DLAGated DeltaNet-1.3B · In-context retrieval averageaccuracy: 20.3 -> 28.8Improved
replace log-linear with DLAMamba-2-780M · RULER 4K S-NIAH-3accuracy: 10.6 -> 37.1Improved
add capacity-bounded memory modeling on top of dynamic state merging (ablation)DLA(I) on Mamba-2 · RULERRULER score: 44.0 -> 57.8Improved
add capacity-bounded memory modeling on top of dynamic state merging (ablation)DLA(I) on Gated DeltaNet · RULERRULER score: 58.4 -> 63.1Improved
add information-aware dynamic merging alone (DLA(I)) versus log-linear (ablation)Mamba-2-780M · RULERRULER score: 39.0 -> 44.0Improved
change cache capacity k from 30 to 20 or 40 (ablation)DLA on Mamba-2 · RULERRULER score: 57.8 -> 56.3 (k=20); 57.7 (k=40)No clear effect
change merge boundary tau from 0.6 to 0.5 or 0.7 (ablation)DLA on Mamba-2 · LAMBADA and RULERaccuracy / RULER score: LAMBADA 23.9 -> 20.1 (0.5), 22.5 (0.7); RULER 57.8 -> 57.3, 55.9Mixed
use DLA instead of log-linear for efficiencyMamba-2-780M · Prefill throughput and memorythroughput and memoryImproved

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