Tiny Models are the Computational Saver for Large Models

Open paper
Benchmark Model Rank Results
image-classification-on-imagenetTinySaver(ConvNeXtV2_h, 0.01 Acc drop)#159Top 1 Accuracy: 86.24GFLOPs: 31.17
image-classification-on-imagenetTinySaver(ConvNeXtV2_h, 0.5 Acc drop)#196Top 1 Accuracy: 85.75GFLOPs: 19.41
image-classification-on-imagenetTinySaver(Swin_large, 0.5 Acc drop)#197Top 1 Accuracy: 85.74
image-classification-on-imagenetTinySaver(Swin_large, 1.0 Acc drop)#239Top 1 Accuracy: 85.24
image-classification-on-imagenetTinySaver(EfficientFormerV2_l, 0.01 Acc drop)#407Top 1 Accuracy: 83.52

Rank counts only results with a code link.