AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

SNP: Structured Neuron-level Pruning to Preserve Attention Scores

KyungHwan Shim, Jaewoong Yun, Shinkook Choi

ECCV 2024 · Architectures and efficiency › Attention and transformers · approach: Vision transformer pruning · 4 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
prune query/key layers with SNP and value layers for inter-head redundancy on DeiT-SmallDeiT-Small · ImageNet-1KJetson Nano latency (ms): 99.32 -> 32.03Improved
use large pruned DeiT-Small with SNP instead of hand-crafted DeiT-TinyDeiT-Small · ImageNet-1KTop-1 accuracy (%): 72.20 -> 73.32Improved
combine SNP with head pruning on DeiT-Base (ablation)DeiT-Base · ImageNet-1KRTX 3090 latency (ms): 72.96 -> 39.31Improved
combine traditional neuron-level masking with head pruning (ablation)head pruning · ImageNet-1KlatencyNo clear effect
compute SNP importance scores with more images (1 to 256) (ablation)SNP · DeiT-Tiny, 70% pruning, no fine-tuningTop-1 accuracy (%): 38.68 -> 42.70Improved
use SNP criterion instead of l2-norm, geometric median or Reverse for query/key pruning (ablation)DeiT-Tiny · ImageNet-1K, no fine-tuningTop-1 accuracyImproved
apply SNP versus WDPruning on DeiT-BaseDeiT-Base · ImageNet-1KGPU speedup: 1.18x -> 2.07xImproved
apply SNP to DeiT-Tiny on GPU versus WDPruningDeiT-Tiny · ImageNet-1KGPU speedup: 1.19x -> 1.05xHurt
convert DeiT MSA split/reshape order for SNP (ablation)DeiT MSA modules · RTX 3090 latencylatency: slower by 0.72 ms (Tiny), 0.43 ms (Base)No 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