Allow to pass different dimension variable order on vision benchmark #60
Description
Given the different backends dimension order, we should be able to prepare the tensors on HxWxC formatting or CxHxW formatting.
{{ message }}
Given the different backends dimension order, we should be able to prepare the tensors on HxWxC formatting or CxHxW formatting.