Hello!
I’m trying to use dlpackrs (https://github.com/ehsanmok/dlpackrs) to convert tvm's NDArray output into ndarray's Array3<f32>
Could you provide a sample code for this?
(context: I’m benchmarking Yolov5s model converted to TVM.
Currently, inference part only takes several microseconds, but output.to_vec() takes 50~90 milliseconds because it copies output from gpu → cpu.)
Thank you!