|
task,model,GPU,batch size,with torch.compile,without torch.compile
|
|
image classification,ViT,V100,1,15.786,16.52
|
|
image classification,BeiT,V100,1,15.71,16.464
|
|
image classification,ConvNeXT,V100,1,30.845,32.932
|
|
image classification,ResNet,V100,1,4.321,6.031
|
|
object detection,DETR,V100,1,53.99,73.1
|
|
object detection,OwlViT,V100,1,51.105,53.634
|
|
object detection,Resnet-101,V100,1,82.566,106.622
|
|
object detection,Conditional-DETR,V100,1,56.984,77.594
|
|
image segmentation,Segformer,V100,1,14.205,16.116
|
|
image segmentation,Mask2former,V100,1,66.815,79.192
|
|
image segmentation,Maskformer,V100,1,188.268,200.026
|
|
image segmentation,MobileNet,V100,1,11.997,18.908
|
|
image classification,ViT,V100,4,43.626,43.653
|
|
image classification,BeiT,V100,4,51.354,52.007
|
|
image classification,ConvNeXT,V100,4,105.58,119.259
|
|
image classification,ResNet,V100,4,11.388,13.039
|
|
object detection,DETR,V100,4,268.003,277.85
|
|
object detection,Resnet-101,V100,4,357.614,421.859
|
|
object detection,Conditional-DETR,V100,4,226.945,289.002
|
|
image segmentation,Segformer,V100,4,42.445,45.327
|
|
image segmentation,Mask2former,V100,4,184.67,201.54
|
|
image segmentation,Maskformer,V100,4,711.28,764.052
|
|
image segmentation,MobileNet,V100,4,48.677,74.289
|
|
image classification,ViT,V100,16,160.907,163.914
|
|
image classification,BeiT,V100,16,187.976,188.978
|
|
image classification,ConvNeXT,V100,16,388.078,422.886
|
|
image classification,ResNet,V100,16,37.604,44.114
|
|
object detection,DETR,V100,16,OOM,OOM
|
|
object detection,Resnet-101,V100,16,1262.758,1619.505
|
|
object detection,Conditional-DETR,V100,16,897.39,1137.513
|
|
image segmentation,Segformer,V100,16,163.62,192.412
|
|
image segmentation,Mask2former,V100,16,695.291,756.337
|
|
image segmentation,Maskformer,V100,16,2656.88,2842.94
|
|
image segmentation,MobileNet,V100,16,201.942,299.003 |