List your company
ResearchFormerly WizWand

Image Classification on Tiny-ImageNet (Average Accuracy)

70.3Average AccuracyQLS
65.46466.719567.97569.2305Jun 23, 2026
Updated 4mo ago

Evaluation results

MethodLinks
QLSBackbone=ResNet-50, Epochs=100, Weight Bitwidth (b)=62026.0670.3
QLS → LSQ(1 epoch)Backbone=ResNet-50, Epochs=101, Weight Bitwidth (b)=62026.0670.28
QLS → LSQ(1 epoch)Backbone=ResNet-50, Epochs=101, Weight Bitwidth (b)=42026.0670.27
QLS → LSQ(1 epoch)Backbone=ResNet-50, Epochs=101, Weight Bitwidth (b)=52026.0670.26
QLSBackbone=ResNet-50, Epochs=100, Weight Bitwidth (b)=52026.0670.14
LSQBackbone=ResNet-50, Epochs=100, Weight Bitwidth (b)=62026.0670.09
QLSBackbone=ResNet-50, Epochs=100, Weight Bitwidth (b)=42026.0670.01
LSQ → LSQ(1 epoch)Backbone=ResNet-50, Epochs=101, Weight Bitwidth (b)=62026.0669.95
LSQBackbone=ResNet-50, Epochs=100, Weight Bitwidth (b)=52026.0669.14
LSQ → LSQ(1 epoch)Backbone=ResNet-50, Epochs=101, Weight Bitwidth (b)=52026.0669.06
LSQBackbone=ResNet-50, Epochs=100, Weight Bitwidth (b)=42026.0667.47
LSQ → LSQ(1 epoch)Backbone=ResNet-50, Epochs=101, Weight Bitwidth (b)=42026.0667.37
QLSBackbone=ResNet-18, Epochs=100, Weight Bitwidth (b)=62026.0666.7
QLSBackbone=ResNet-18, Epochs=100, Weight Bitwidth (b)=52026.0666.24
LSQBackbone=ResNet-18, Epochs=100, Weight Bitwidth (b)=62026.0666.2
QLSBackbone=ResNet-18, Epochs=100, Weight Bitwidth (b)=42026.0665.82
LSQBackbone=ResNet-18, Epochs=100, Weight Bitwidth (b)=52026.0665.8
LSQBackbone=ResNet-18, Epochs=100, Weight Bitwidth (b)=42026.0665.65