List your company
ResearchFormerly WizWand

Image Classification on CIFAR-10 8000 images (test)

93.2AccuracyV1b var
21.9640.45558.9577.445May 10, 2026
Updated 6mo ago

Evaluation results

MethodLinks
V1b varSparsity (S)=0.5, Backbone=ViT-Small, Feature class=κ=0 / κ=1 class, Statistical protocol=3 seeds2026.0593.2
V1 4-smt.Sparsity (S)=0.5, Backbone=ViT-Small, Feature class=κ=0 / κ=1 class, Statistical protocol=3 seeds2026.0593.2
A5 4-rawSparsity (S)=0.5, Backbone=ViT-Small, Feature class=κ=2 class, Statistical protocol=3 seeds2026.0593.2
V1a peakSparsity (S)=0.5, Backbone=ViT-Small, Feature class=κ=0 / κ=1 class, Statistical protocol=3 seeds2026.0592.7
NL_sinSparsity (S)=0.5, Backbone=ViT-Small, Feature class=control, Statistical protocol=3 seeds2026.0592.6
PureMag4Sparsity (S)=0.5, Backbone=ViT-Small, Feature class=control, Statistical protocol=3 seeds2026.0592.6
RandSpl.Sparsity (S)=0.5, Backbone=ViT-Small, Feature class=κ=0 / κ=1 class, Statistical protocol=3 seeds2026.0592.5
A5_Log6Sparsity (S)=0.5, Backbone=ViT-Small, Feature class=κ=2 class, Statistical protocol=3 seeds2026.0592
Taylor baselineSparsity (S)=0.5, Backbone=ViT-Small, Feature class=Baseline, Statistical protocol=3 seeds2026.0591.1
V1 4-smt.Sparsity (S)=0.6, Backbone=ViT-Small, Feature class=κ=0 / κ=1 class, Statistical protocol=3 seeds2026.0586.8
RandSpl.Sparsity (S)=0.6, Backbone=ViT-Small, Feature class=κ=0 / κ=1 class, Statistical protocol=3 seeds2026.0586.7
V1a peakSparsity (S)=0.6, Backbone=ViT-Small, Feature class=κ=0 / κ=1 class, Statistical protocol=3 seeds2026.0586.5
V1b varSparsity (S)=0.6, Backbone=ViT-Small, Feature class=κ=0 / κ=1 class, Statistical protocol=3 seeds2026.0585.8
A5_Log6Sparsity (S)=0.6, Backbone=ViT-Small, Feature class=κ=2 class, Statistical protocol=3 seeds2026.0585.7
A5 4-rawSparsity (S)=0.6, Backbone=ViT-Small, Feature class=κ=2 class, Statistical protocol=3 seeds2026.0585.1
NL_sinSparsity (S)=0.6, Backbone=ViT-Small, Feature class=control, Statistical protocol=3 seeds2026.0583.9
PureMag4Sparsity (S)=0.6, Backbone=ViT-Small, Feature class=control, Statistical protocol=3 seeds2026.0583.4
Taylor baselineSparsity (S)=0.6, Backbone=ViT-Small, Feature class=Baseline, Statistical protocol=3 seeds2026.0575.7
A5 4-rawSparsity (S)=0.7, Backbone=ViT-Small, Feature class=κ=2 class, Statistical protocol=3 seeds2026.0569.3
V1 4-smt.Sparsity (S)=0.7, Backbone=ViT-Small, Feature class=κ=0 / κ=1 class, Statistical protocol=3 seeds2026.0568.8
V1b varSparsity (S)=0.7, Backbone=ViT-Small, Feature class=κ=0 / κ=1 class, Statistical protocol=3 seeds2026.0566.8
NL_sinSparsity (S)=0.7, Backbone=ViT-Small, Feature class=control, Statistical protocol=3 seeds2026.0564.8
A5_Log6Sparsity (S)=0.7, Backbone=ViT-Small, Feature class=κ=2 class, Statistical protocol=3 seeds2026.0562.9
RandSpl.Sparsity (S)=0.7, Backbone=ViT-Small, Feature class=κ=0 / κ=1 class, Statistical protocol=3 seeds2026.0562.7
V1a peakSparsity (S)=0.7, Backbone=ViT-Small, Feature class=κ=0 / κ=1 class, Statistical protocol=3 seeds2026.0562.2
Taylor baselineSparsity (S)=0.7, Backbone=ViT-Small, Feature class=Baseline, Statistical protocol=3 seeds2026.0560
PureMag4Sparsity (S)=0.7, Backbone=ViT-Small, Feature class=control, Statistical protocol=3 seeds2026.0558
A5 4-rawSparsity (S)=0.8, Backbone=ViT-Small, Feature class=κ=2 class, Statistical protocol=3 seeds2026.0540.3
V1a peakSparsity (S)=0.8, Backbone=ViT-Small, Feature class=κ=0 / κ=1 class, Statistical protocol=3 seeds2026.0537.7
RandSpl.Sparsity (S)=0.8, Backbone=ViT-Small, Feature class=κ=0 / κ=1 class, Statistical protocol=3 seeds2026.0537.2
V1 4-smt.Sparsity (S)=0.8, Backbone=ViT-Small, Feature class=κ=0 / κ=1 class, Statistical protocol=3 seeds2026.0535.6
NL_sinSparsity (S)=0.8, Backbone=ViT-Small, Feature class=control, Statistical protocol=3 seeds2026.0535.5
A5_Log6Sparsity (S)=0.8, Backbone=ViT-Small, Feature class=κ=2 class, Statistical protocol=3 seeds2026.0534.2
PureMag4Sparsity (S)=0.8, Backbone=ViT-Small, Feature class=control, Statistical protocol=3 seeds2026.0533.9
V1b varSparsity (S)=0.8, Backbone=ViT-Small, Feature class=κ=0 / κ=1 class, Statistical protocol=3 seeds2026.0532.7
Taylor baselineSparsity (S)=0.8, Backbone=ViT-Small, Feature class=Baseline, Statistical protocol=3 seeds2026.0524.7