Is there any way to split a deep learning model into several devices?
E.g:
resnet50 --order=[dla1,gpu,dla0] --partition-point-1=13 --partition-point-2=25
Where, order is the order of the devices in the pipeline and the partition points correspond to the model segments in layers to be mapped onto each of the declared devices. I don’t know if there is such utility in either deepstream or tensor_rt. I would like to use something like that on a jetson agx xavier.