Problem: model_transmute_deploy inference speed is lesser than model_transmute.
model_transmute_deploy size is 12.2 kb
model_transmute size is 36.2 kb
Multiple input tensor size experimented with : [1,3,16,256,256], [1,3,4,256,256], [1,3,16,6,6]
model_transmute_deploy inference time: 5.375,1.102, 0.0627sec
model_transmute inference time:0.1049,0.0107,0.0245 sec
provide the reason for this is to why its this happening and also the optimal solutions to overcome this issue
pytorch video accelerator script link: Accelerate your model with model transmuter in PytorchVideo/Accelerator ยท PyTorchVideo