List your company
ResearchFormerly WizWand

Pruning on CIFAR-100 (test)

0.669FLOPs-based Accuracy-Retention AUCBest local-axis method
0.085560.237030.38850.53997May 8, 2026
Updated 5mo ago

Evaluation results

MethodLinks
Best local-axis methodBackbone=ResNet-182026.050.669
Random channel pruningBackbone=ResNet-182026.050.614
Best target-axis methodBackbone=ResNet-182026.050.606
MagnitudeBackbone=VGG-162026.050.563
FPGM / geometric medianBackbone=VGG-162026.050.561
Network SlimmingBackbone=ResNet-182026.050.546
Network SlimmingBackbone=VGG-162026.050.525
Random channel pruningBackbone=VGG-162026.050.523
Best local-axis methodBackbone=VGG-162026.050.483
TaylorBackbone=VGG-162026.050.482
Best target-axis methodBackbone=VGG-162026.050.448
Best local-axis methodBackbone=MobileNetV22026.050.42
FPGM / geometric medianBackbone=ResNet-182026.050.396
MagnitudeBackbone=ResNet-182026.050.393
TaylorBackbone=ResNet-182026.050.349
Best target-axis methodBackbone=MobileNetV22026.050.346
TaylorBackbone=MobileNetV22026.050.329
Activation RMSBackbone=MobileNetV22026.050.32
FPGM / geometric medianBackbone=MobileNetV22026.050.315
MagnitudeBackbone=MobileNetV22026.050.313
Random channel pruningBackbone=MobileNetV22026.050.285
CHIPBackbone=MobileNetV22026.050.284
Network SlimmingBackbone=MobileNetV22026.050.267
Activation RMSBackbone=ResNet-182026.050.243
Activation RMSBackbone=VGG-162026.050.14
CHIPBackbone=ResNet-182026.050.13
CHIPBackbone=VGG-162026.050.108