|
task,model,GPU,batch size,with torch.compile,without torch.compile
|
|
image classification,ViT,A100,1,7.584,9.325
|
|
image classification,BeiT,A100,1,8.448,11.282
|
|
image classification,ConvNeXT,A100,1,10.208,10.41
|
|
image classification,ResNet,A100,1,4.124,6.531
|
|
object detection,DETR,A100,1,19.04,34.619
|
|
object detection,OwlViT,A100,1,18.42,24.978
|
|
object detection,Resnet-101,A100,1,18.197,36.276
|
|
object detection,Conditional-DETR,A100,1,17.993,31.219
|
|
image segmentation,Segformer,A100,1,10.5,11.759
|
|
image segmentation,Mask2former,A100,1,49.117,60.188
|
|
image segmentation,Maskformer,A100,1,59.487,75.764
|
|
image segmentation,MobileNet,A100,1,3.974,8.583
|
|
image classification,ViT,A100,4,14.499,14.832
|
|
image classification,BeiT,A100,4,13.048,13.205
|
|
image classification,ConvNeXT,A100,4,21.631,22.94
|
|
image classification,ResNet,A100,4,4.268,6.657
|
|
object detection,DETR,A100,4,32.418,48.657
|
|
object detection,Resnet-101,A100,4,44.998,68.101
|
|
object detection,Conditional-DETR,A100,4,35.552,56.47
|
|
image segmentation,Segformer,A100,4,16.476,18.838
|
|
image segmentation,Mask2former,A100,4,61.781,74.277
|
|
image segmentation,Maskformer,A100,4,159.116,180.7
|
|
image segmentation,MobileNet,A100,4,8.515,14.174
|
|
image classification,ViT,A100,16,40.01,40.944
|
|
image classification,BeiT,A100,16,41.048,41.854
|
|
image classification,ConvNeXT,A100,16,75.561,82.258
|
|
image classification,ResNet,A100,16,5.024,7.018
|
|
object detection,DETR,A100,16,161.902,164.382
|
|
object detection,Resnet-101,A100,16,155.021,232.887
|
|
object detection,Conditional-DETR,A100,16,124.032,180.491
|
|
image segmentation,Segformer,A100,16,31.144,37.005
|
|
image segmentation,Mask2former,A100,16,154.814,178.945
|
|
image segmentation,Maskformer,A100,16,579.826,638.57
|
|
image segmentation,MobileNet,A100,16,30.31,51.693 |