AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Combiner: Full Attention Transformer with Sparse Computation Cost

Hongyu Ren, Hanjun Dai, Zihang Dai, Mengjiao Yang, Jure Leskovec, Dale Schuurmans, Bo Dai

NeurIPS 2021 · Architectures and efficiency › Attention and transformers · approach: Sparse efficient transformers for long context · 28 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 sparse attention with Combiner full attention using the same axial pattern (ablation)6-layer image density models · CIFAR-10Bits/Dim: 3.666 -> 3.050Improved
replace sparse attention with Combiner full attention using the same fixed pattern (ablation)6-layer image density models · CIFAR-10Bits/Dim: 3.408 -> 3.321Improved
replace sparse attention with Combiner full attention using the same Logsparse pattern (ablation)6-layer image density models · CIFAR-10Bits/Dim: 4.253 -> 3.366Improved
add a second partition plan through a mixture (ablation)Combiner-Axial · CIFAR-10, 6-layerBits/Dim: 3.050 -> 3.040Improved
replace sparse attention with Combiner full attention using the same axial pattern (ablation)6-layer image density models · ImageNet-64Bits/Dim: 4.032 -> 3.585Improved
replace sparse attention with Combiner full attention using the same fixed pattern (ablation)6-layer image density models · ImageNet-64Bits/Dim: 3.696 -> 3.654Improved
replace sparse attention with Combiner full attention using the same Logsparse pattern (ablation)6-layer image density models · ImageNet-64Bits/Dim: 4.351 -> 3.795Improved
add a second partition plan through a mixture (ablation)Combiner-Axial · ImageNet-64, 6-layerBits/Dim: 3.585 -> 3.585No clear effect
use DeepSets-Max instead of max pooling for abstraction summaries (ablation)Combiner · Wiki-40B, 8kperplexity: 16.60 -> 16.29Improved
use DeepSets-Mean instead of mean pooling for abstraction summaries (ablation)Combiner · Wiki-40B, 8kperplexity: 16.54 -> 16.48Improved
use mean pooling instead of max pooling for abstraction summaries (ablation)Combiner · Wiki-40B, 8kperplexity: 16.60 -> 16.54No clear effect
increase context length from 2k to 8kCombiner-Fixed · Wiki-40Bperplexity: 18.04 -> 16.60Improved

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