AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Generalizable Multi-linear Attention Network

Tao Jin, Zhou Zhao

NeurIPS 2021 · Architectures and efficiency › Attention and transformers · approach: Sparse and structured attention · 3 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
remove combinational addition operation (CAO) (ablation)MAN (Non-Transformer) · CMU-MOSIBA: 79.4 -> 78.4Improved
remove combinational addition operation (CAO) (ablation)MAN (Transformer) · CMU-MOSIBA: 82.7 -> 81.5Improved
remove local sequential constraints (LSC) (ablation)MAN (Transformer) · CMU-MOSIBA: 82.7 -> 81.4Improved
remove hierarchical approximation decomposition (HAD) (ablation)MAN (Non-Transformer) · CMU-MOSIBA: 79.4 -> 79.4No clear effect
remove hierarchical approximation decomposition (HAD) (ablation)MAN (Transformer) · CMU-MOSIBA: 82.7 -> 82.6No clear effect
remove HAD (high-order attention map) (ablation)MAN · CMU-MOSIFLOPs: 3.87 × 105 -> 10.23 × 106Hurt
remove CAO (ablation)MAN · CMU-MOSIFLOPs: 3.87 × 105 -> 8.31 × 106Hurt
remove LSC (ablation)MAN · CMU-MOSIFLOPs: 3.87 × 105 -> 3.07 × 105Improved
use Transformer backbone instead of non-TransformerMAN · CMU-MOSIBA: 79.4 -> 82.7Improved
add MAN module to LMF backbone, compare to LMFMAN (Non-Transformer) · CMU-MOSIMAE: 0.912 -> 0.894Improved
add MAN module to MulT backbone, compare to MulTMAN (Transformer) · CMU-MOSIBA / F1: 82.8 -> 83.0Mixed
add MAN module to MMT backbone, compare to MMTMAN (Transformer) · MSR-VTT text-to-videoR@1: 24.6 -> 24.1No clear effect

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