AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Fast Monte-Carlo Approximation of the Attention Mechanism

Hyunjun Kim, JeongGil Ko

AAAI 2022 · Architectures and efficiency › Attention and transformers · approach: Attention head and block redesign · 6 citations (OpenAlex)

Read the paper

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
Use attention-weighted Monte-Carlo sampling with α=0.2BERT · SST-2accuracy: 92.43 -> 92.26No clear effect
Use attention-weighted Monte-Carlo sampling with α=0.2BERT · GLUE (CoLA)FLOPS reduction: 11.44$\times$Improved
Increase the error coefficient α from 0.2 to 1.0 (ablation)MCA-BERT · SST-2FLOPS reduction: 5.58$\times$ -> 12.34$\times$Improved
Increase the error coefficient α from 0.2 to 1.0 (ablation)MCA-BERT · SST-2accuracy: 92.26$\pm$0.0 -> 80.66$\pm$0.1Hurt
Increase the error coefficient α from 0.2 to 1.0 (ablation)MCA-BERT · CoLAMatthews correlation: 53.74$\pm$0.1 -> 50.95$\pm$0.7Hurt
Apply Monte-Carlo Attention with α=0.2DistilBERT · SST-2accuracy: 91.51 -> 91.50No clear effect
Apply Monte-Carlo Attention with α=0.2Longformer · AAPDaccuracy: 74.81 -> 74.70No clear effect
Apply Monte-Carlo Attention with α=0.2Longformer · HNDFLOPS reduction: 5.32$\times$Improved

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