AI Paper Atlas

Research map / Architectures and efficiency / Mixture of experts

ShiftAddViT: Mixture of Multiplication Primitives Towards Efficient Vision Transformer

Haoran You, Huihong Shi, Yipin Guo, Yingyan Celine Lin

NeurIPS 2023 · Architectures and efficiency › Mixture of experts · approach: Sparse MoE as regularization and attention · 5 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 dense attention MatMuls with additions and use shifts/MoE for linear and MLP layersPVTv2-B0 · ImageNetlatency: 7.82 -> 1.51 msImproved
Replace dense attention MatMuls with additions and use shifts/MoE for linear and MLP layersPVTv2-B0 · ImageNetaccuracy: 70.44 -> 70.59Improved
Replace attention and MLP operations with additions, shifts, and MoEGNT · LLFF averaged, eight sceneslatency: 1071 -> 746.6Improved
Replace MLPs with Shift layers rather than using MoE (ablation)PVTv1-T · ImageNet sensitivity analysisaccuracy: 76.21 -> 73.92Hurt
Use MoE instead of Shift-only MLPs (ablation)PVTv1-T · ImageNet sensitivity analysisaccuracy: 73.92 -> 74.81Improved
Use vanilla Q/K quantization instead of KSH (ablation)ShiftAddViT · ImageNet classificationaccuracy: ↓0.05% ~ ↑0.21%Mixed
Add latency-aware load-balancing loss to MoE routing (ablation)ShiftAddViT · PVTv2-B0 and PVTv1-Tnormalized latency: 100% -> 85.4%; 100% -> 85.5%Improved
Add latency-aware load-balancing loss to MoE routing (ablation)ShiftAddViT · PVTv2-B0 and PVTv1-Taccuracy: 70.38 -> 70.37; 74.73 -> 74.66No clear effect
Replace linear/MLP layers with shift layers (ablation)ShiftAddViT · LLFF novel view synthesisPSNR: ↑0.13 ~ ↑0.20Improved
Use linear attention instead of MSA (ablation)PVTv2-B0 · ImageNet, batch size 1, input resolution 224latency: 4.62 -> 6.34 msHurt

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