AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Paramixer: Parameterizing Mixing Links in Sparse Factors Works Better Than Dot-Product Self-Attention

Tong Yu, Ruslan Khalitov, Lei Cheng, Zhirong Yang

CVPR 2022 · Architectures and efficiency › Attention and transformers · approach: Sparse and structured attention · 11 citations (OpenAlex)

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
Parameterize mixing links in sparse full-rank factors instead of using dot-product attentionParamixerNet · Long Range Arena Textaccuracy: 68.4% -> 83.32%Improved
Use CHORD rather than CDIL sparse-link protocol (ablation)Paramixer · Long Range Arena Pathfinderaccuracy: 67.13±0.42 -> 80.49±0.13Improved
Use CDIL rather than CHORD sparse-link protocol (ablation)Paramixer · Long Range Arena Imageaccuracy: 45.01±0.21 -> 46.58±0.05Improved
Use CDIL rather than CHORD sparse-link protocol (ablation)Paramixer · Long Range Arena Textaccuracy: 83.12±0.33 -> 83.32±0.19No clear effect
Use CHORD rather than CDIL sparse-link protocol (ablation)Paramixer · Long Range Arena ListOpsaccuracy: 37.78±0.28 -> 39.57±0.32Improved
Parameterize mixing links from the original input rather than the preceding block's output (ablation)ParamixerNet · Paramixer neural-network experimentstraining performanceImproved
Use longer context (32k rather than 16k tokens) (ablation)Paramixer · Long Documentaccuracy: 83.89 -> 84.55Improved
Parameterize sparse-factor mixing links instead of dot-product attentionParamixer · HFDNAROC AUC: 94.32 -> 100.00Improved
Parameterize sparse-factor mixing links instead of dot-product attentionParamixer · MTcDNAROC AUC: 79.38 -> 84.86Improved

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